Adjusting filter or classification control settings

Abstract

Methods and systems for managing data communications are described. The method includes receiving a data communication; analyzing the data communication to determine a particular type of sender or recipient activity associated with the data communication based at least in part on an application of a plurality of tests to the data communication; assigning a total risk level to the data communication based at least in part on one or more risks associated with the particular type of sender or recipient activity and a tolerance for each of the one or more risks; comparing the total risk level assigned to the data communication with a maximum total acceptable level of risk; and allowing the data communication to be delivered to a recipient in response to the comparison indicating that the total risk level assigned to the data communication does not exceed the maximum total acceptable level of risk.

Description

TECHNICAL FIELD

This document relates generally to systems and methods for processing communications and more particularly to systems and methods for classifying and/or filtering communications.

BACKGROUND

Internet connectivity has become central to many daily activities. For example, millions of people in the United States use the internet for various bill pay and banking functionalities. Countless more people use the internet for shopping, entertainment, to obtain news, and for myriad other purposes. Moreover, many businesses rely on the internet for communicating with suppliers and customers, as well as providing a resource library for their employees. However, various entities still use the internet for malicious or non-reputable activity. For example, spammers send billions of messages daily causing headaches for many users and administrators.

Reputation systems and message profilers have enabled better recognition of non-reputable traffic. Reputation systems and message profilers can take a one size fits all approach to identifying non-reputable entities and/or messages. Such an approach can lead to users opting out of such protection and ignoring the information provided by such systems.

SUMMARY

In one aspect, systems, methods, apparatuses and computer program products are provided. In one aspect, methods are disclosed, which comprise: receiving a data communication in a network; analyzing the data communication to determine a particular type of sender or recipient activity associated with the data communication based at least in part on an application of a plurality of tests to the data communication; assigning a total risk level to the data communication based at least in part on one or more risks associated with the particular type of sender or recipient activity and a tolerance for each of the one or more risks; comparing the total risk level assigned to the data communication with a maximum total acceptable level of risk; and allowing the data communication to be delivered to a recipient in response to the comparison indicating that the total risk level assigned to the data communication does not exceed the maximum total acceptable level of risk.

Systems can include a security control interface, a policy control interface and a filtering module. The security control interface can produce security control representations, each of the security control representations being operable to control security settings associated with a protected entity. The policy control interface can produce policy control representations, each of the policy control representations being operable to control policy settings associated with a protected entity. The filtering module can filter one or more communication streams based upon the plurality of security settings and based upon the plurality of policy settings.

Computer readable media can include program code operable to enable adjustment of filter and/or classification settings for incoming and outgoing communications, causing a processor to performs steps including: receiving a plurality of ranges from an administrator; providing a security control interface to a user, the security control interface comprising a plurality of security control representations associated with a plurality of security control settings, each of the security control mechanisms including an associated range from among the plurality of ranges, the associated range defining a minimum and maximum setting associated with the respective security controls; receiving input from the user through the security control interface, the input requesting adjustment of the security control settings; adjusting a plurality of thresholds related to plurality of control settings received from the user, the plurality of thresholds being associated with tolerance for a classification of potential security violation; and, filtering communications streams from a protected entity associated with the user based upon the plurality of thresholds.

DESCRIPTION OF DRAWINGS

FIG. 1A is a block diagram of an example network environment including a security agent.

FIG. 1B is a block diagram of another example network architecture including local security agents.

FIG. 2 is a block diagram of an example security agent.

FIG. 3A is a block diagram of an example central reputation module

FIG. 3B is a block diagram of an example security agent.

FIG. 4 is a block diagram of a system providing resolution between a global reputation and a local reputation.

FIG. 5 is a screen shot of an example graphical user interface used to adjust the settings of a communications filter.

FIG. 6 is a screen shot of another example graphical user interface used to adjusting the settings of a communications filter.

FIG. 7 is a block diagram of another example interface used to adjust the settings of a communications filter.

FIG. 8 is a block diagram of another example interface used to adjust the settings of a communications filter.

FIG. 9 is a flowchart of an example method for adjusting filter settings associated with a network security agent.

DETAILED DESCRIPTION

FIG. 1A is a block diagram of an example network environment including a security agent. In various implementations, the network environment can include a security agent 100, a protected network 110 including a number of agents 120a-f, and a plurality of external entities 130a-f connected via an external network 140. The security system, in various implementations, can include a reputation system 150 and/or a message profiler 160. In some implementations, the security agent 100 can reside between a firewall system (not shown) and a network 110 (e.g., an enterprise network). The network 110, in various implementations, can include a plurality of agents (e.g., computers). In further implementations, the network 110 can include a number of servers, including, for example, electronic mail servers, web servers, and various application servers as may be used by the enterprise associated with the network 110.

In various implementations, the security agent 100 can monitor communications entering and exiting the network 110. These communications can be received, for example, through an external network 120 (e.g., the Internet) from any of a number of entities 130a-f connected to the external network 120. One or more of the entities 130a-f can be legitimate originators of communications traffic while other(s) of the entities 130a-f can also be non-reputable entities originating unwanted communications. However, it can be difficult to know in advance which of the entities 130a-f are originating unwanted communications and which are originating legitimate communications. As such, in some implementations, the security agent 100 can include a reputation engine 150.

In various implementations, the reputation engine can inspect a communication and to determine a reputation of an entity 130a-f associated with the communication. In some implementations, the security agent 100 can determine what action to take with the communication based upon the reputation of the originating entity. For example, if the reputation indicates that the originator of the communication is reputable the security agent can forward the communication to the recipient of the communication. However, if the reputation indicates that the originator of the communication is non-reputable, for example, the security agent can quarantine the communication, perform more tests on the message, or require authentication from the message originator, among many others. Reputation engines are described in detail in United States Patent Publication No. 2006/0015942, which is hereby incorporated by reference.

In further implementations, the security agent 100 can monitor incoming communications and derive the reputable and non-reputable characteristics of a communication by identifying the characteristics associated with the communication. For example, the attributes associated with communications can be compared to the attributes of known reputable and known non-reputable messages to determine whether the communication is legitimate or not. Message profiling is describe in detail in U.S. application Ser. No. 11/173,941, entitled “Message Profiling Systems and Methods,” filed on Jun. 2, 2005.

In some implementations, a tolerance for specific types of messages or activity can be set by a user via an agent 120a-f or an administrator of the security agent 100. However, it can be impractical for a user to set tolerances for each individual attribute. Thus, in some implementations, the user can provide control settings through an agent 120a-f for general classifications of activities. For example, a user associated with one of agents 120a-f might have a high tolerance for spam messages, and a low tolerance for virus activity. Thus, the security agent 100 can be set to include a low threshold for attributes or senders that might be originating virus content, while also being set to include a high threshold for attributes or senders that might be originating spam content.

In some implementations, a low threshold for classifying a communication can lead to overinclusive classification of a category, while a high threshold for classifying a communication can lead to an underinclusive classification of a category. For example, when a threshold is set low, a communication does not need to exhibit many of the attributes associated with a classification before being assigned to the classification. Likewise, when the threshold is set high, a communication would exhibit many of the attributes associated with a classification before being assigned to the classification. In some implementations, when the threshold is set high, positive attributes associated with the communication can be overweighted in the analysis. In other implementations, when the threshold is set low, positive attributes associated with the communication can be underweighted in the analysis.

FIG. 1B is a block diagram illustrating an example network architecture including local security agents. In the example of FIG. 2, a number of agents 170a-c can be coupled to an external network 175. In some implementations, the agents 170a-c can reside on personal computing devices (e.g., personal computer, mobile communications device, laptop computer, personal digital assistant, etc). The agents 170a-c can include a local reputation engine and/or a message profiler.

A number of other entities 180a-c can also be coupled to the network 175. The entities 180a-c can send communications on the network. Some of the communications originating from the entities 180a-c can be directed to personal computing devices that include agents 170a-c. The agents 170a-c can receive the communications and determine what action to take with the communication based upon the reputation of the sending entity 180a-c or a profile (e.g., attributes) associated with the communication. For example, if the receiving agent 170a-c determines that the message is not legitimate, or that the sending entity 180a-c is not legitimate, the agent 170a-c can determine to delay delivery of the communication, place the communication in a quarantine, mark the communication, or drop the communication, among others. In other examples, where the message is legitimate, the agent 170a-c can allow a program associated with the communication to access the communication.

In some implementations, the agents 170a-c can include a user interface. The user interface can allow users associated with the agents 170a-c to adjust control setting associated with the agents 170a-c. For example, if the user has a particular aversion to bulk communications (e.g., spam messages), the user can set the control settings to classify communications as bulk even where the communication only shares a few characteristics with known bulk communications. Similarly, if the user is not concerned about bulk messages, and is instead more concerned with misclassification of communications, the user can set the control settings to provide a stricter classification of bulk communications, thereby identifying fewer communications as bulk communications. In another example, the user and/or administrator might have a particular aversion to phishing websites. As such, the user can set the control settings to classify communications as associated with phishing websites even where the communications only shares a few characteristics with known communications associated with phishing websites.

In some implementations, the user might not care about the type of content that is being delivered but rather a communications risk associated with that content. In such implementations, the user can set the control settings to provide a higher tolerance for various content classifications associated with incoming data (e.g., including e-mail, web-data, instant messaging data, voice over internet protocol (VoIP) data, or any other application used to communicate data) presented and provide a lower tolerance for classification of communications risk associated with any incoming data.

In some implementations, an optional central reputation server 185 can be included. The central reputation server 185 can provide reputation information regarding the network entities 180a-c to the agents 170a-c upon request. For example, one of the agents 170a-c might receive a communication from an entity 185a-c from which it has previously not received communications. As such, the agent 170a-c might not be configured to determine a reputation associated with the entity 180a-c. However, the agent 170a-c can send a reputation request to a central reputation server 185 to determine a reputation associated with the entity 180a-c.

In further implementations, the agents 170a-c can outsource all reputation functions to the central reputation server 185. Thus, each communication received by the agents 170a-c can be queried to the central reputation server 185. In further implementations, the agents 170a-c can provide control settings to the central reputation server 185. In such implementations, the central reputation server can use the control settings received from the agents 170a-c to bias the respective query responses sent to each of the agents 170a-c.

In some implementations, the central reputation server can be a reputation server associated with a large network receiving and/or routing a large volume of traffic daily. The central reputation server in such instances could provide a reputation based upon the communications and behavior that the central reputation server has directly observed. In further implementations, the central reputation server 185 could be one of many central reputation servers configured to provide reputation information to agents 170a-c.

In other implementations, the central reputation server can provide an aggregation of multiple local reputations, such as the local reputations derived by agents 170a-c. For example, the central reputation server might periodically poll a plurality of agents 170a-c for reputation information associated with all the traffic the agents 170a-c have observed. The reputation server 185 can then aggregate the local reputation information to provide a global reputation.

FIG. 2 is a block diagram of an example security agent 200. In various implementations, the security agent 200 can receive data from a user or an external entity 210 (e.g., depending on the direction of the data flow) at an interface module 220. The communications can be received through any of a variety of communications media (e.g., wired or wireless).

In some implementations, the interface module can provide the communications to a reputation and/or classification engine 230. The reputation and/or classification module 230 can operate to provide reputation of entities associated with data and/or classification of data received. In some implementations, the reputation and/or classification module 230 can be a part of the security agent 200. In other implementations, the reputation and/or classification module 230 can be provided to the security agent 200 from a central device.

In various implementations, reputation can be derived based upon one or more entities associated with various types of data. The past activities of such entities can be used to predict the future activities of these entities. For example, if an entity originates spam with a high frequency, there is some likelihood that future communications include spam. In some implementations, as the frequency with which entities engage in an activity rises, the likelihood that the future activity exhibits the same characteristic also rises. In further implementations, the reputation can also be time dependent. For example, an entity might have malware (e.g., a “bot”) installed on the device. Such malware can periodically send data to other devices at predetermined times. Such entities can be assigned reputations that include time dependencies. In those implementations including a reputation engine, such characteristics can be identified by the reputation and/or classification module 230 and provided to the security agent 200 for use in determining whether to allow the associated data.

In some implementations, classification of data can be derived based upon the similarities between data content and previously classified content. For example, a user/administrator can identify a corpus of documents associated with a particular classification. The corpus of documents have been identified by the user/administrator as exhibiting the characteristics belonging to the particular classification identified. A classification engine can analyze the corpus of documents to identify similarities between the documents, thereby identifying the defining characteristics of the particular classification of documents. Subsequently received data can then be tested to determine whether the data exhibits the same features as the corpus of documents characteristic of the particular classification. In various implementations, multiple classifications of documents can be identified by providing multiple sets of document corpora. For example, a user/administrator might provide a document corpus for a spam classification, a document corpus for a phishing classification, a document corpus for a malware classification, or a document corpus for a spyware classification, among many others. In those implementations including a classification engine, such classification characteristics can be identified by the reputation and/or classification module 230 and provided to the security agent 200 for use in determining whether to allow the associated data.

In some implementations, the security agent 200 can include an risk assessment module 240. The risk assessment module 240 can operate to assess a risk associated with the particular activity 240 being engaged in by the user. In various implementations, risks can include a spectrum of activities. The closer an activity lies to an origin of a risk scale associated with the activity, the safer the activity is. The further away an activity is from the origin of the risk scale associated with the activity, the more risk involved in the activity. In various examples, risk classifications can include: liability, maturity, network security, bandwidth exhaustion, communication freedom, information security, and low productivity. Other risk classifications can be used. The risk assessment module 240 can assess the activity to identify the risk associated with the activity.

In various implementations, the security agent 200 can include a filter module 250. The filter module 250 can receive an assessed risk associated with an activity from the risk assessment module 240. The filter module 250 can apply a policy to determine whether the activity falls within an acceptable risk level. In those instances where the activity is outside of an acceptable risk level, the activity can be blocked and an administrator and/or user 260 can be notified. In those instances where the activity is within an acceptable risk level, the activity can be allowed and the user 260 can be allowed to continue the activity.

In various implementations, the security agent can also include a control settings interface 270. The control settings interface 270 can facilitate the provision of policy by a user and/or administrator 260. The control settings interface 270 can allow the user and/or administrator to tune his/her own comfort levels with respect to risk associated with activities engaged in by the user(s). For example, the classification of data and reputation of the entity originating the data can be conflated using multiple transform into several risk classifications. An assessment of the risk associated with the data in these different categories can then be made by comparing the risk associated with the activity to a policy to determine whether the activity is prohibited by policy.

FIG. 3A is a block diagram of an example central reputation module. In some implementations, control settings can be used to adjust the determination of a global reputation by the central reputation module based on local reputation feedback. A security agent 300 can send a query through a network 310 to a reputation module 320. In some implementations, the reputation module 320 can be local to the security agent 300. In some examples, the security agent 300 originates a query in response to receiving a communication from an unknown entity. Alternatively, the security agent 300 can originate the query responsive to receiving any communications, thereby promoting use of more up-to-date reputation information.

The reputation module 320 is operable to respond to the query with a global reputation determination. The central reputation module 320 can derive the reputation using a reputation engine 330. In some implementations, the reputation engine 330 is operable to receive a plurality of local reputations 332 from a respective plurality of local reputation engines. In some examples, the plurality of local reputations 332 can be periodically sent by the reputation engines to the reputation module 320. Alternatively, the plurality of local reputations 332 can be retrieved by the server upon receiving a query from a security agent 300. In some implementations, the reputation module 330 can be a central reputation module and the plurality of local reputations 332 can include a local reputation derived by a local reputation engine associated with the security agent 300.

The local reputations can be combined using confidence values 334 related to each of the local reputation engines and then accumulating the results. The confidence value 334 can indicate the confidence associated with a local reputation produced by an associated reputation engine. Reputation engines associated with individuals, for example, can receive a lower weighting in the global reputation determination. In contrast, local reputations associated with reputation engines operating on large networks can receive greater weight in the reputation determination based upon the confidence value 334 associated with that reputation engine.

In some implementations, the confidence values 334 can be based upon feedback received from users. For example, a reputation engine that receives a lot of feedback indicating that communications were not properly handled because local reputation information 332 associated with the communication indicated the wrong action can be assigned low confidence values 334 for local reputations 332 associated with those reputation engines. Similarly, reputation engines that receive feedback indicating that the communications were handled correctly based upon local reputation information 332 associated with the communication indicated the correct action can be assigned a high confidence value 334 for local reputations 332 associated with the reputation engine.

In some implementations, a tuner 340 can be used to adjust the confidence values associated with the various reputation engines. The tuner can to receive control settings 342 and can adjust the confidence values based upon the received control settings 342. For example, the agent 300 can provide control settings 342 including a variety of security settings, such as any of virus sensitivity settings, computer worm sensitivity settings, phishing detection sensitivity settings, trojan horse sensitivity settings, among many other types of malware sensitivities and combinations thereof

In other examples, the agent 300 can provide control settings that include a variety of policy settings, such as spyware sensitivity settings, spam sensitivity settings, bulk mail sensitivity settings, content sensitivity settings, user privacy sensitivity settings, and many other policy settings to protect the network from unwanted or illegal activity, including, in some instances, communications traffic originating from the protected computer (e.g., preventing spyware infecting a computer from providing keystroke logs to external entities). In some implementations, various settings might be classified as security settings, while other settings might be reclassified as policy settings. For example, spam settings can be classified under security settings and/or policy settings depending on a configuration desired by an administrator or user. In other examples, phishing settings can be classified as a security setting or as a policy setting based upon the configuration provided by an administrator or a user.

In various implementations, the confidence values 334 can be provided to the reputation module 320 by the local reputation engine itself based upon stored statistics for incorrectly classified entities. In other implementations, information used to weight the local reputation information can be communicated to the reputation module 320.

In some examples, a bias 345 can be applied to the resulting global reputation vector. The bias 345 can normalize the reputation vector to provide a normalized global reputation vector to a security agent 300. Alternatively, the bias 345 can be applied to account for local preferences associated with the security agent 300 originating the reputation query. Thus, a security agent 300 can receive a global reputation vector matching the defined preferences of the querying security agent 300. The security agent 300 can take an action on the communication based upon the global reputation vector received from the reputation module 320.

In some implementations, the preferences used in the bias 345 can include control settings 342 provided by the security agent 300 to the central reputation module 320. For example, the agent 300 can provide control settings 342 including a variety of security settings and/or policy settings. In some implementations, security settings can include policy settings. In industry, policy settings are often set up as security measures to prevent misuse of company equipment and to secure sensitive trade secrets from exposure.

FIG. 3B is a block diagram of an example security agent 350. In some implementations, control settings can be used to adjust the classification and/or filtering of messages by a the security agent 350. In various implementations, the security agent can be configured to examine an existing set of related files 392, as specified by an administrator 394, to create identification characteristics associated with the set of related files 392. The files 392 can be supplied to a classification engine 355. In various implementations, the classification engine 355 might use any of the classification techniques 375, 380, 385 on the set of related files 392 to determine the features or characterizations identifying relationships between the files 392. For example, legal documents might include Latin phrases such as in re, ipso facto, or prima facie. Such an inclusion could be discoverable by a message classification engine 355.

In some implementations, a classification engine 355 can generate a set of identifying characteristics for a class of content. The same techniques 375, 380, 385 are then used on communications entering the security agent 350. In these implementations, the characteristics of the communication may then be compared to the identifying characteristics for a class of content to determine in which class (if any) the content of the communication belongs, thereby producing a content classification 390 for the communication. In some implementations, the security agent 350 can apply a policy related to the content classification, as shown by the filtering module 360, to determine whether the communication will be delivered via network 365, or dropped, quarantined, etc. as shown by block 370. In the event that a communication does not satisfy policy, the originating system 396 can be alerted to the policy failure. The messaging content compliance system could also notify a system user or administrator 394 and/or sender of the policy failure.

In some implementations, the security agent 350 can provide a tuner interface 398 to a user or an administrator 394. The tuner interface can be used by the user or administrator to make adjustments to the classification engine 355. In some implementations, the user or administrator 394 can provide control settings to the classification engine 355. The control settings can adjust the level of similarity found between received messages and classification characterizations before classifying the received message as belonging to a classification. In other implementations, the control settings can adjust the range into which a message characteristics may fall before being classified as belonging to a classification. In some implementations, the provision of control settings can dynamically adjust the level of filtering associated with the security agent. In other implementations, the provision of control settings can adjust the level of system resources consumed by the security agent.

In still further implementations, the tuner interface 398 can provide an interface that facilitates adjustment of the filter module 360 based upon the type of traffic. For example, the administrator/user might determine that electronic mail should be monitored more closely than other types of data communications (e.g., web traffic, music/video streaming traffic, instant messaging traffic, etc.). In such examples, the administrator can use the interface to adjust the control settings based upon the type of traffic being filtered.

In other implementations, the tuner interface 398 can provide an interface to facilitate adjustment of risk tolerance associated with communications. In such implementations, an administrator and/or user can provide control settings to the filter indicating one or more classifications of traffic that the filter should pass based upon the risk associated with the communications. Thus, while data might be loosely or strictly classified by the classification engine 355 the filter module 360 can be adjusted by the tuner interface to provide strict enforcement of filter policies or loose enforcement of filter policies based upon the risk associated with data of the various classifications.

FIG. 4 is a block diagram illustrating an example resolution between a global reputation and a local reputation. In some implementations, a security agent (e.g., security agent 300 of FIG. 3A) can use global reputation data and provide resolution between global reputation data and local reputation data. The security agent 400 communicates with a server 420 to retrieve global reputation information from the server 420. The local security agent 400 can receive a communication at 402. The local security agent can correlate the communication to identify attributes of the message at 404. The attributes of the message can include, for example, an originating entity, a fingerprint of the message content, a message size, etc. The local security agent 400 includes this information in a query to the server 420. In other examples, the local security agent 400 can forward the entire message to the server 420, and the server can perform the correlation and analysis of the message.

The server 420 uses the information received from the query to determine a global reputation based upon a configuration 425 of the server 420. The configuration 425 can include a plurality of reputation information, including both information indicating that a queried entity is non-reputable 430 and information indicating that a queried entity is reputable 435. The configuration 425 can also apply a weighting 440 to each of the aggregated reputations 430, 435. A reputation score determinator 445 can provide the engine for weighting 440 the aggregated reputation information 430, 435 and producing a global reputation vector.

The local security agent 400 then sends a query to a local reputation engine at 406. The local reputation engine 408 performs a determination of the local reputation and returns a local reputation vector at 410. The local security agent 400 also receives a response to the reputation query sent to the server 420 in the form of a global reputation vector. The local security agent 400 then mixes the local and global reputation vectors together at 412. An action is then taken with respect to the received message at 414.

In some implementations, the mixer 412 can include security and/or policy control settings received from the recipient 450. For example, a recipient 450 associated with an adult content provider might not be interested in blocking communications including adult content from the computer. Thus, the recipient 450 might raise a threshold associated with classifying a particular communication as non-reputable based upon adult content included in the communication. In various implementations, multiple security and/or policy control settings can be provided. For example, such settings can include, malware detection settings, virus detection settings, phishing detection settings, trojan horse detection settings, logic bomb detection settings, zombie detection settings, spyware detections settings, click fraud detection settings, distributed denial of service detection settings, spam detection settings, bulk communications detection settings, policy violation detection settings, among many others, and combinations thereof.

FIG. 5 is an example graphical user interface 500 for adjusting the settings of a filter associated with a reputation server. The graphical user interface 500 can allow the user of a local security agent to adjust the settings of a local filter in several different categories 510, such as, for example, “Virus,” “Worms,” “Trojan Horse,” “Phishing,” “Spyware,” “Spam,” “Content,” and “Bulk.” However, the categories 510 depicted are merely examples. Other categories are possible.

In various implementations, this disclosure provides an interface that enables users and/or administrators to dynamically control filtering of inbound and outbound network traffic. For example, a user and/or administrator can adjust the filtering to lower spam filtering but adjust the filtering to raise porn filtering. In additional implementations, the user and/or administrator can “tune” the amount of inbound and outbound filtering. For example, a network administrator can be responsible for the performance of the network as well as the identification of unwanted network traffic. Thus, the network administrator might want to increase throughput of a filtering system to boost network performance, while sacrificing the identification of some types of traffic.

In some implementations, the categories 510 can be divided into two or more types of categories. For example, the categories 510 of FIG. 5 are divided into a “Security Settings” type 520 of category 510, and a “Policy Settings” type 530 of category. Other categories are possible. In each of the categories 510 and types 520, 530, a mixer bar representation 540 can allow the user to adjust the particular filter setting associated with the respective category 510 of communications or entity reputations.

Moreover, while categories 510 of “Policy Settings” type 530 in various implementations can be adjusted freely based upon the user's own judgment, categories of “Security Settings” type 520 can be limited to adjustment within a range. This limitations on adjustment can be set in order to prevent a user from altering the security settings of the security agent beyond a range acceptable to a network administrator. For example, a disgruntled employee could attempt to lower the security settings, thereby leaving an enterprise network vulnerable to attack. Thus, in some implementations, the range limitations 550 placed on categories 510 in the “Security Settings” type 520 are operable to keep security at a minimum level to prevent the network from being compromised. In some implementations, the ranges 550 can be hard-wired by the system. In other implementations, the ranges 550 can be adjusted by the administrator or based upon a profile selected by a user and/or administrator. Such ranges 550 can prevent users from subverting network settings.

In various examples, the “Policy Settings” type 530 categories 510 are those types of categories 510 that would not compromise the security of a network, but might only inconvenience the user or the enterprise if the settings were lowered. Thus, in some implementations the “Policy Settings” 530 can be adjusted freely. In other implementations, the “Policy Settings” 530 can include limitations set by the administrator, by a profile, or hard-wired by the system.

Furthermore, in some implementations, range limits 550 can be placed upon all of the categories 510. Thus, the local security agent could prevent users from setting the mixer bar representation 540 outside of the provided range 550. In additional implementations, the ranges may not be shown on the graphical user interface 500. Instead, the range 550 could be abstracted out of the graphical user interface 500 and all of the settings would be relative settings. Thus, the category 510 could display and appear to allow a full range of settings, while transforming the setting into a setting within the provided range. For example, the “Virus” category 510 range 550 is provided in this example as being between level markers 8 and 13. If the graphical user interface 500 were set to abstract the allowable range 550 out of the graphical user interface 500, the “Virus” category 510 would allow setting of the mixer bar representation 540 anywhere between 0 and 14. However, the graphical user interface 500 could transform the 0-14 setting to a setting within the 8 to 13 range 550. Thus, if a user requested a setting of midway between 0 and 14, the graphical user interface could transform that setting into a setting of midway between 8 and 13.

In some implementations, control setting ranges can be limited by system performance. For example, if a system has a given processing power, the system might not have the capacity to provide for certain control settings. In some implementations, a system performance control setting can be provided. The system performance control setting can be used to provide an administrator or user with a representation of the system performance of the network based on the given security and policy control settings. For example, if each of the security and policy control settings is adjusted to a maximum level, the system performance control setting might be at lower than a maximum level.

In some implementations, adjustment of the system performance setting can cause the security and/or policy control settings to be adjusted. For example, if the security and policy control settings are set at the highest levels, and the administrator attempts to adjust the system performance setting to the maximum level, the interface can reduce the security and policy control settings to provide for maximum performance. In some implementations, the ratio between the settings can be maintained when the system performance control setting is adjusted.

In some implementations, a user can create his/her own categories for classifications. In such implementations, a tuner interface can allow the user or administrator to provide a catalog of files (e.g., documents, messages, etc.) that are representative of a group of documents for which the user intends to create a classification. A message classification engine can examine the documents and derive any relationships between the documents and derive the characteristics associated with the group of documents. For example, a user might determine that all unsolicited traffic is unwanted. Such a user can create a single classification for unsolicited traffic, which can comprise any of a number of classifications associated with the and provide those documents that belong to the classification to a message classification engine. The message classification engine can thereby create a classification and allow the user to tune the tolerance associated with the classification.

FIG. 6 is another example graphical user interface 600 for adjusting the settings of a filter associated with a reputation server. In some implementations, the graphical user interface 600 can allow the user to customize the settings view. For example, in the user interface 600 a plurality of sliders 640 can be moved horizontally to adjust policy control settings 620 and/or security control settings 630. The categories 610 can be displayed to provide the user with a list of categories corresponding to each of the sliders 640. In some implementations, the sliders can be limited to stay within a range 650 provided, for example, by an administrator.

In other implementations, the graphical user interface might provide knob representations, operable to provide the user with an interface for tuning security and/or policy control settings by turning the knob representations. Other interface abstractions are possible.

FIG. 7 is a block diagram of another example interface 700 used to adjust the settings of a communications filter. In some implementations, the interface 700 can include a number of risk classifications 710. In some examples, the risk classifications can include, among others: liability, maturity, network security, bandwidth exhaustion, communication freedom, information security, low productivity and general information.

The interface 700 can include a scale representation 720 against which the risk classifications 710 can be adjusted. In some implementations, adjusting a control setting to high tolerance on the scale 720 can allow more types of information associated with that activity category to be accessed, while adjusting the control setting to low tolerance on the scale 720 can prevent users from engaging in activities that might include some level of risk to an associated network.

In some implementations, the scale 720 can also include a blacklist setting representation 730 and a whitelist setting representation 740. The blacklist setting representation 730 can provide more freedom to users by allowing access to any types of activities associated with a risk classification 710 except where the activity is prohibited by a blacklist entry. The whitelist setting representation 740 can provide more limited access to activities associated with a risk classification 710 by blocking all activities associated with that risk classification 710 except where the activity is included on a whitelist associated with the risk classification 710.

In some implementations, the interface 700 can include a number of control setting representations 750a-h. The control setting representations 750a-h can represent risk classifications 710 including liability, maturity, network security, bandwidth exhaustion, communication freedom, information security, low productivity and general information, respectively. The various control settings can enable a user or administrator to adjust control settings associated with the respectively risk classification 710.

In some implementations, a liability risk classification setting 750a can define a user's risk tolerance with respect to legal liability. In various instances, an employee might engage in network activity could subject an employer to legal liability. For example, if a user is sending adult content to other employees, the employee may be subjecting the employer to liability for sexual harassment charges on the part of the recipient. In other examples, the employee may be engaging in criminal activities that could subject the employer to liability, such as defrauding customers under the color of his duties to the employer or embezzling money.

In some implementations, a maturity risk classification setting 750b can include content that may harm the emotional development of children. For example, in environments where children may be present, an administrator might want to limit access to content that contains mature content. In other examples, an environment such as a workplace might not include children. Such workplace environments can provide higher tolerance to mature content. In various examples, mature content can include (in no particular order) such concepts as violence, gambling, drugs, profanity, or nudity, among many others.

In some implementations, network security risk classification settings 750c can include filtering content such as sources of malware that might damage computer software, circumvent filtering policy or expose confidential data. Any network activity can expose network assets to security risks. However, a goal of a network is not to block all network traffic, but to limit the risks associated with the activities engaged in by users.

In some implementations, a bandwidth exhaustion risk classification setting 750d can include filtering risks that an application may exhaust the bandwidth of the network. Some applications, such as for example, streaming media, can use an inordinate amount of bandwidth and can cause greater risk that the network may not have enough bandwidth to serve new applications. For example, if a business is served by a digital subscriber line (DSL) connection, multiple streaming media connections (such as, e.g., streaming video, web phone, etc.) may increase the risk that an important teleconference connection cannot be established. In other examples, a business might have a T3 network connection. In such examples, bandwidth is probably less of a concern for an administrator or user based upon the available bandwidth. As such, an user/administrator can track the usage of the network and set the risk levels appropriately based upon historical usage.

In some implementations, communication freedom risk classification setting 750e can include filtering the types of applications with which users can communicate with other users. For example, many web mail programs and web phone are difficult to manage risk because they allow the user to communicate information directly to other users outside of the organization. Thus, some administrators can provide limits to the types of applications that can be used by network entities.

In some implementations, information security risk classification setting 750f can include filtering risks that a user or application exposes confidential company information to users outside of the company. For example, peer-to-peer file sharing can allow users outside of a business access to files on a business computer. In other examples, phishing websites and/or applications can allow external users to receive data associated with employees of a company, and even allow access to business information.

In some implementations, productivity risk classification setting 750g can include filtering risks that employees are using the network resources for non-productive uses. Many companies do not wish to limit access to such resources for fear that employees would feel untrusted and not develop loyalty to the company. However, many websites and/or applications can reduce the productivity of a company's employees. Thus, the company might want to reduce total use of such websites and/or applications, while not eliminating access to these sites and/or applications.

In some implementations, general information risk classification setting 750h can include filtering of fact based sites that allow the user to review information about various subjects. If an administrator sets the control setting for the general information category at a low tolerance, the policy engine will limit the information accessed to those categories of information useful for their job, rather than information on a location of the user's polling place or science websites describing why the sky is blue. If the control settings are adjusted to high tolerance (or blacklist), the policy engine will allow the user to access most types of general information. So, if there is a site with lots of information that has some risk that might be blocked by your Bandwidth or Propriety setting, over-ride those other risks to let this site through.

In various implementations, other risk classifications can be used.

FIG. 8 is a block diagram of another example interface used to adjust the settings of a communications filter. In various implementations, a number of risk classifications can be identified. Risk classifications can be mapped onto a multi-dimensional risk space 800. The multi-dimensional risk space can be used, in some implementations, to aid a user or administrator in determining how to set and apply policy for an associated network (or a network device).

In some implementations, each of the risk classifications can be represented by an axis 810-870 and can be identified as orthogonal (or substantially orthogonal) to each other. In such implementations, each of the axes 810-870 associated with the risk classification is substantially independent of the other axes and can be envisioned to lie at right angles to the other axes. The axes 810-870 can be used to map activities based upon the risk associated with that activity to a respective risk classification. For example, illegal activity can imply liability issues for a company. In some instances, an activity can imply multiple risk classifications. For example, an employee viewing porn might imply a liability risk and a maturity risk if there are young people that might be affected by the content (such as a children's television production company).

The multi-dimensional risk space 800 can also include a number of control settings 750a-g. In some implementations, the multi-dimensional risk space can be provided as a companion interface for a slider interface (e.g., interface 700 of FIG. 7). In these implementations, the control settings 750a-g can provide an indication of the types of activity being filtered by the various settings made with respect to the slider interface.

In other implementations, the multi-dimensional risk space 800 can be provided as a primary control setting interface for a filtering module (e.g., filtering module 250 of FIG. 2). In such implementations, an administrator and/or user can adjust the control settings 750a-g by moving the control setting 750a-g away from an origin or toward an origin of the respective axis 810-870 associated with the control setting 750a-g. For example, if an administrator places a high priority on network security, the administrator could move a network security control setting 750c toward the origin of a network security axis 830. In other examples, if an administrator and/or user place a low priority on network bandwidth (e.g., because there is plenty of available bandwidth), the administrator and/or user could move a bandwidth control setting 750d away from the origin of a bandwidth axis 840. Other axes can be used.

In still further implementations, total risk can be calculated based upon an area associated with an activity. For example, based upon the multi-dimensional space 800, it can be identified that there might exist an activity that involves more than two different risks. In such instances, a filter can compute an area defined by the activity on the risk classification axes 810-870 and calculate a total risk associated with the activity based upon the risks implicated by the activity on each of the implicated risk classification axes 810-870. In such implementations, a user and/or administrator can define a maximum total level of risk acceptable to the user and/or administrator, and the calculated total risk of an activity can be compared to the maximum total level of acceptable risk to determine whether to allow an activity.

FIG. 9 is a flowchart illustrating an example method for adjusting filter settings associated with a network security agent. At stage 900, ranges are identified. The ranges can be identified, for example, by a security agent (e.g., security agent 100 of FIG. 1). In various implementations, the identified ranges can provide a range within which the control setting can be adjusted. In some implementations, the range can be depicted in a user interface environment. In other implementations, the control setting can be a weighting associated with a classification category within a range of weightings, whereby the setting comprises a percentage operator (e.g., multiplier) upon a default setting. In some examples, an administrator can be allowed to set a maximum multiplier and a minimum multiplier, and the maximum and minimum multipliers comprise the range within which the control setting can be adjusted. In still further implementations, the control setting can be a direct adjustment of a threshold associated with a classification and the range can be a reflection of a minimum and maximum threshold associated with the classification. In some implementations, the range can be hard coded into the security agent. In some such implementations, an administrator can be provided authorization to provide further limitations within the range allowed by the security agent.

At stage 910, a control interface is provided. The control interface can be provided, for example, by a security agent (e.g., security agent 100 of FIG. 1) in conjunction with a user interface. In some implementations, the control interface can include a plurality of sliders operable to adjust control settings associated with a communications filter. For example, the control interface can provide an equalizer representation to the user, enabling the user to adjust a plurality of detection settings for various classifications of communications that can be filtered by a communications filter.

At stage 920, control settings can be received. The control settings can be received, for example, by a security agent (e.g., security agent 100 of FIG. 1) in conjunction with a user interface. In some implementations, the control settings are received based upon a user accessing the control interface and adjusting a control representation associated with the control interface. In various implementations, the control settings can provide direct adjustment of classification thresholds. In other implementations, the control settings can operate as a percentage operator on a default bias associated with a classification. For example, a default setting could represent the identity of a default bias, while adjusting the control setting higher or lower could multiply or divide the default bias, respectively.

At stage 930, thresholds associated with a type of communication can be adjusted. The thresholds can be adjusted, for example, by a security agent (e.g., security agent 100 of FIG. 1) in conjunction with a user interface. In some implementations, the thresholds associated with a type of communication can define the level of similarity required between a subject communication and communications previously associated with the type before a subject communication will be classified as belonging to that communication type. For example, a high threshold might require many similarities between the subject communication and classified communications before classifying the subject communication as belonging to the type. Similarly, a low threshold might require only a few similarities between the subject communication and previously received communications of the classification type before classifying the subject communication as belonging to the type. In various implementations, the similarities between communications can be derived based upon multiple attributes associated with the communication, including similarities to known legitimate communications.

In other implementations, stage 930 can be replaced by a stage operable to adjust a bias associated with a type of communication. For example, the weightings associated with specific attributes or characteristics of a subject can be adjusted to provide results to a profiler. The weightings can cause a profiler to associate a communication with a particular classification based upon certain attributes of the communication being amplified with respect to other attributes.

At stage 940, communications can be filtered based on thresholds. The communications can be filtered, for example, by a security agent (e.g., security agent 100 of FIG. 1), in conjunction with a communications filter. In some implementations, the filter can drop communications of certain classifications. In other implementations, the filter can quarantine communications for further inspection. In still further implementations, the filter can mark communications based upon a classification. In some such implementations, the markings associated with a communication can be made available to the user and/or programs associated with the user's computer.

In alternative implementations, stage 940 can be replaced by filtering the communications based on classifications identified by results of the profiler, and as biased based on the control settings. For example, a user might request that all communications be marked based upon a classification, or that certain classifications be dropped, quarantined, or otherwise delayed.

The systems and methods disclosed herein may use data signals conveyed using networks (e.g., local area network, wide area network, internet, etc.), fiber optic medium, carrier waves, wireless networks (e.g., wireless local area networks, wireless metropolitan area networks, cellular networks, etc.), etc. for communication with one or more data processing devices (e.g., mobile devices). The data signals can carry any or all of the data disclosed herein that is provided to or from a device.

The methods and systems described herein may be implemented on many different types of processing devices by program code comprising program instructions that are executable by one or more processors. The software program instructions may include source code, object code, machine code, or any other stored data that is operable to cause a processing system to perform methods described herein.

The systems and methods may be provided on many different types of computer-readable media including computer storage mechanisms (e.g., CD-ROM, diskette, RAM, flash memory, computer's hard drive, etc.) that contain instructions for use in execution by a processor to perform the methods' operations and implement the systems described herein.

The computer components, software modules, functions and data structures described herein may be connected directly or indirectly to each other in order to allow the flow of data needed for their operations. It is also noted that software instructions or a module can be implemented for example as a subroutine unit of code, or as a software function unit of code, or as an object (as in an object-oriented paradigm), or as an applet, or in a computer script language, or as another type of computer code or firmware. The software components and/or functionality may be located on a single device or distributed across multiple devices depending upon the situation at hand.

This written description sets forth the best mode of the invention and provides examples to describe the invention and to enable a person of ordinary skill in the art to make and use the invention. This written description does not limit the invention to the precise terms set forth. Thus, while the invention has been described in detail with reference to the examples set forth above, those of ordinary skill in the art may effect alterations, modifications and variations to the examples without departing from the scope of the invention.

As used in the description herein and throughout the claims that follow, the meaning of “a,” “an,” and “the” includes plural reference unless the context clearly dictates otherwise. Also, as used in the description herein and throughout the claims that follow, the meaning of “in” includes “in” and “on” unless the context clearly dictates otherwise. Finally, as used in the description herein and throughout the claims that follow, the meanings of “and” and “or” include both the conjunctive and disjunctive and may be used interchangeably unless the context clearly dictates otherwise.

Ranges may be expressed herein as from “about” one particular value, and/or to “about” another particular value. When such a range is expressed, another embodiment includes from the one particular value and/or to the other particular value. Similarly, when values are expressed as approximations, by use of the antecedent “about,” it will be understood that the particular value forms another embodiment. It will be further understood that the endpoints of each of the ranges are significant both in relation to the other endpoint, and independently of the other endpoint.

These and other implementations are within the scope of the following claims.

Claims

1. A system comprising: one or more hardware data processors; andinstructions stored on a computer readable storage medium operable, when executed by the one or more hardware data processors, to: receive a data communication in a network;analyze the data communication to determine a particular type of sender or recipient activity associated with the data communication based at least in part on an application of a plurality of tests to the data communication;assign a total risk level to the data communication based at least in part on one or more risks associated with the particular type of sender or recipient activity and a tolerance for each of the one or more risks;compare the total risk level assigned to the data communication with a maximum total acceptable level of risk; andallow the data communication to be delivered to a recipient in response to the comparison indicating that the total risk level assigned to the data communication does not exceed the maximum total acceptable level of risk.

2. The system of claim 1, wherein the tolerance for each of the one or more risks is set in a security policy for the network.

3. The system of claim 1, wherein the tolerance for each of the one or more risks is adjustable along a tolerance range from a high tolerance setting to a low tolerance setting, the high tolerance setting specifying fewer restrictions on a delivery of a data communication to a recipient than the low tolerance setting.

4. The system of claim 3, wherein the tolerance range comprises a blacklist setting, the blacklist setting prohibiting a particular type of sender or recipient activity.

5. The system of claim 3, wherein the tolerance range comprises a whitelist setting, the whitelist setting allowing a particular type of sender or recipient activity.

6. The system of claim 3, wherein the instructions are further operable, when executed by the one or more processors, to: accept tolerance limits on the tolerance range from a network administrator limiting the tolerance range to a range less than an entire tolerance range; andrestrict a setting of the tolerance for each of the one or more risks based on the tolerance limits.

7. The system of claim 1, wherein the one or more risks are associated with one or more of a liability category, a maturity category, a security category, bandwidth category, a communications category, an information security category, or a productivity category.

8. A method comprising: receiving a data communication in a network;analyzing the data communication to determine a particular type of sender or recipient activity associated with the data communication based at least in part on an application of a plurality of tests to the data communication;assigning a total risk level to the data communication based at least in part on one or more risks associated with the particular type of sender or recipient activity and a tolerance for each of the one or more risks;comparing the total risk level assigned to the data communication with a maximum total acceptable level of risk; andallowing the data communication to be delivered to a recipient in response to the comparison indicating that the total risk level assigned to the data communication does not exceed the maximum total acceptable level of risk.

9. The method of claim 8, wherein the tolerance for each of the one or more risks is set in a security policy for the network.

10. The method of claim 8, wherein the tolerance for each of the one or more risks is adjustable along a tolerance range from a high tolerance setting to a low tolerance setting, the high tolerance setting specifying fewer restrictions on a delivery of a data communication to a recipient than the low tolerance setting.

11. The method of claim 10, wherein the tolerance range comprises a blacklist setting, the blacklist setting prohibiting a particular type of sender or recipient activity.

12. The method of claim 10, wherein the tolerance range comprises a whitelist setting, the whitelist setting allowing a particular type of sender or recipient activity.

13. The method of claim 10, further comprising: receiving tolerance limits on the tolerance range from a network administrator limiting the tolerance range to a range less than an entire tolerance range; andrestricting a setting of the tolerance for each of the one or more risks based on the tolerance limits.

14. The method of claim 8, wherein the one or more risks are associated with one or more of a liability category, a maturity category, a security category, bandwidth category, a communications category, an information security category, or a productivity category.

15. A non-transitory computer-readable storage medium storing instructions that are executable by one or more processing devices, and upon execution, cause the one or more processing devices to perform processes comprising: receiving a data communication in a network;analyzing the data communication to determine a particular type of sender or recipient activity associated with the data communication based at least in part on an application of a plurality of tests to the data communication;assigning a total risk level to the data communication based at least in part on one or more risks associated with the particular type of sender or recipient activity and a tolerance for each of the one or more risks;comparing the total risk level assigned to the data communication with a maximum total acceptable level of risk; andallowing the data communication to be delivered to a recipient in response to the comparison indicating that the total risk level assigned to the data communication does not exceed the maximum total acceptable level of risk.

16. The non-transitory computer readable medium of claim 15, wherein the tolerance for each of the one or more risks is set in a security policy for the network.

17. The non-transitory computer readable medium of claim 15, wherein the tolerance for each of the one or more risks is adjustable along a tolerance range from a high tolerance setting to a low tolerance setting, the high tolerance setting specifying fewer restrictions on a delivery of a data communication to a recipient than the low tolerance setting.

18. The non-transitory computer readable medium of claim 17, wherein the tolerance range comprises a blacklist setting, the blacklist setting prohibiting a particular type of sender or recipient activity.

19. The non-transitory computer readable medium of claim 17, wherein the tolerance range comprises a whitelist setting, the whitelist setting allowing a particular type of sender or recipient activity.

20. The non-transitory computer readable medium of claim 17, wherein the instructions cause the one or more processing devices to further perform processes comprising: receiving tolerance limits on the tolerance range from a network administrator limiting the tolerance range to a range less than an entire tolerance range; andrestricting a setting of the tolerance for each of the one or more risks based on the tolerance limits.