Patent Application
20050100157
This invention relates generally to communication systems, and more particularly to a context aware call processing architecture for effecting user-defined features.
Traditional telephony features are standardized and coarse grained. Users are given access to a plurality of features (Call Forward on Busy etc.) that are centrally located within a communication system, such as a PBX.
The utility of these features is limited by their inability to be personalized to the needs of an individual and to be aware of the user's current activities. For example, a user may be occupied with some work. Such a user will therefore be ‘busy’ in relation to other work of less importance but ‘available’ in relation to work of higher priority or to information about his/her current task. Traditional telephony features such as Call Forward on Busy are coarse-grained and therefore cannot match these important human distinctions.
The very limited degree of personalization in feature sets according to the prior art results from the fact that traditional call processing has relied on a centralized definition of common features to accelerate the time to market and to improve system stability. Expert designers typically develop standard features that are tested for compatibility before deployment. Personalization with such a scheme is impossible because of the large numbers of possible personalized features.
The Internet telephony work on the SIP protocol (IETF) has created the CPL standard that partially addresses the goal of achieving personalized features (see Sinnreich, H; Johnston, A. B., Internet Communication Using SIP, Wiley, 2001 pp 104-108, and The Internet Draft by Schulzrinne, H.; Rosenberg, J.; SIP Caller Preferences and Callee Capabilities, draft-ietf-sip-callerprefs-05.txt available at: http://www.softarmor.com/sipwg/drafts/draft-ietf-sip-callerprefs-05.txt). Although CPL allows the user a limited ability to locally specify call handling, it does not go beyond specifying policies on call handling based on the identity of the incoming caller and time of day. It does not take into account the user's context and is unable to accommodate conflict handling between user policies, thereby severely limiting the utility of CPL, it's scalability and evolvability.
According to the present invention, an architecture and implementation are provided by which telephony features can be personalized to meet fine-grained needs. The parameters that affect personalization features are considered to be part of a user's work context and so these features are called ‘context-aware’.
More particularly, the system according to the present invention is capable of:
Unique aspects of this invention include:
A detailed description of the invention is set forth herein below, with reference to the following drawings, in which:
Turning to
Our co-pending patent application filed on the same day as this application and entitled “Availability and Location Predictor Using Call Processing Indications”, describes a method of evidence gathering that surmises (i.e. makes guesses) about user availability from his/her interactions with a PBX system. The interactions are gathered as awareness information and processed by an algorithm into availability information. By using awareness information as evidence, the algorithm makes predictions about the users' availability or makes decisions in call processing. This information is fed back into a shared database (e.g. a tuple space) as assertions that indicate high level assessments about the user's state. These assessments are then used by the call processing components described below to make decisions on call handling.
Additional inventive aspects relating to the application of awareness data 1 to context-based communication systems is set forth in our copending applications filed on the same date as this application and entitled “System and method for facilitating communication using presence and communication services” and “Generation of Availability Indicators from Call Control Policies for Presence Enabled Telephony System”.
Awareness data (i.e. raw information about the user) that has been processed by context engine 3 into availability information is then applied to a policy engine 5 in the form of assertions of interest to policies within the system, as described in detail below. The policy engine 5 includes a context update block 9 and feature selection policies 11.
In context update block 9, an incoming event (e.g. Invite, etc.) is related to the current context of the user. Every event has associated with it some indicators that are pertinent to the user's call that provide evidence of the relevance, urgency and importance of the call to the user. Such indicators include caller identity, role relationship between caller and called party, group or project membership, location of user, current state of called user, subject of the call, and so on. Some of these evidential indicators are explicit in the call and some can be derived by inference from other indicators (e.g. awareness data, as discussed above). The context update block 9 uses opportunistic reasoning to develop the required evidence. This evidence is then supplied to the feature selection policies 11 for selecting a feature, discussed in greater detail below. As discussed in our counterpart application entitled “Generation of Availability Indicators from Call Control Policies for Presence Enabled Telephony System”, the evidential indicators may be in the form of fuzzy variables. The fuzziness of these variables is used to indicate the degree of credence that the system has in them.
In the context update evidence gathering process performed by block 9, the system occasionally asks the user which feature he/she wishes to have executed. Sending the user options to his/her wireless browser asking for his/her selection is one among several mutually appropriate options for accomplishing this. Moreover, options may be sent to the calling party asking him/her to select one of several acceptable options.
As discussed above, the user sets some of the policies used in the opportunistic reasoning of context update block 9, although many of the policies are implicit in the way people generally interact. Thus, several default policies are set by system administrators using sociological principles. Examples of such default policies include calls from superiors being more important than calls from subordinates, someone alone in an office being more available than someone with several visitors, calls with a subject related to a user's current activities are less intrusive than calls with unrelated subjects, etc.
Once the user context has been updated with call specific information in block 9, the feature to be performed is selected in block 11. Feature selection policies block 11 utilizes the evidence generated previously in block 9 to direct call handling in accordance with policies set by the user governing the personalized call handling he/she expects. These policies indicate how calls should be handled for differing role relationships, time of day, user state, user location etc. According to applicant's counterpart application filed on the same day as this application and entitled “Personalizable and Customizable Feature Execution and Specification System for IP Telephony Using Operational Semantics and Deontic Task Trees”, block 11 uses forward chaining with fuzzy reasoning to generate priorities among all proposed features and relates that to the intrusiveness of the proposed actions. This results in the selection of a single feature as being the most appropriate for call handling given the declared personalized preferences of the user. The feature is then executed, as shown at block 13.
The execution of features at block 13 can be modulated as described in co-pending UK Application No. 9920051.1 (Call Processing by Synchronized Tuple Spaces) to allow external features acting as enterprise constraints to control the execution of the selected feature.
Thus, with the new addressing capabilities of Internet-based telephony, features can take on new semantics. Instead of remaining at the device level as in conventional telephony, features can operate at a personal level. Calls are not directed to physical endpoints but rather to aspects of the user's identity within his/her business context. A user may have multiple aspects to his/her identity, each of which has different capabilities in the business environment. For example, a user may have multiple aspects to his identity in the form of:
These multiple identities may be carried directly in URLs that convey meaning in respect to aspects of the user's identity using the standard ‘dot’ convention. Thus, a user named John Doe may have multiple aspects of his business identity specified as:
This change in convention effectively creates an entirely new model of how telephony features operate in converged voice and data systems. As indicated above, according to this invention, features operate with an understanding of the current user context and communications are directed to the most appropriate aspect of the user identity. Thus, for example, calling parties can indicate which aspects of the identity they wish to contact and which aspects they do not wish to deal with.
An exemplary hardware implementation of the system according to the present invention is shown in
The ACD client 21 consists of user interfaces and user agents. The user interfaces provide an introduction window, a login window for a registered system user, and a registration window for a new user. Knowledge management is an important part of the user interface on the client system. A user can create or manage personal information such as a buddy list, relationship information, a schedule and a user preference rule. The client server utilizes two types of agents: The Call Delivery (CD) agent and the System Management (SM) agent. The CD agent acknowledges events, which are generated by the Call Monitor, in the Tspaces server 23. The Call Monitor is a direct interface with the MiTAI Gateway 25, and creates an event that is fed into the TSpaces server 23 for starting call processing by the CD agent. Next, the SM agent acknowledges the event from the CD agent, and distributes the call processing to agents on the network. Although each agent has distinct services, both the server and the client have certain common modules as per customary object-oriented design. These common object modules and the other modules are explained below.
The Event Handler subsystem 33 is a monitoring daemon that resides between the user interface 31 and the Application Level subsystem 35. It waits for physical events to arrive from the user interface 31, such as mouse clicks, and directs them to the appropriate application module. The development tool, Java, provides embedded event handlers, such as ActionListener, for this purpose.
The Application Level 35 is the core of the system. It consists of multiple agents that provide services for a client as well as for a server. All of the system transactions, functionalities, and knowledge management are performed within this subsystem.
As shown in the class diagram of
Each agent is also responsible for registering an event into the TSpaces 37 to communicate with client machines. Whenever an awaited tuple is written into the TSpaces, the TSpaces server 23 notifies this to an agent that registered the event. Generating an event and getting notification of the event from the TSpaces 37 forms a two-way communication-acknowledgement between agents.
The Relationship Assigning (RA) agent is responsible for responding to a relationship-assigning request from a client's SM agent. The request from a SM agent contains caller and receiver information. The RA agent assigns the relationship between the user and the caller according to the user's buddy-list.
The User Rule Assigning (URA) agent is responsible for responding to a user-rule-assigning request from a client's SM agent. Upon request, the URA agent retrieves both the relationship information and the user's current contexts. The relationship information is a relationship between the caller and the receiver, set by the RA agent. The user's current contexts are the user's location, the current time with the user's schedule, and the user's activity.
The User-rule Conflict Resolving (UCR) agent is responsible for responding to a client's SM agent for the user-rule conflict-resolving request. The request contains user rule information that is assigned by the URA agent. The UCR agent selects one rule that is the most specific among the assigned rules. The more conditions a rule has, the more specific a rule is considered to be. The Call Simulation service is provided for testing without connecting to a MiTAI Gateway 25. A window with a form serves this functionality.
As shown in the client module class diagram of
The Call Delivery (CD) agent is responsible for communication with the phone switch or PBX 26 through MiTAI Gateway 25. In particular, the CD agent registers events to MiTAI Gateway 25 and waits for the notification of an incoming call for a user. When the notification arrives, the CD agent sends a request to the SM agent for further processing and waits for response. This response from the SM agent contains an action to be taken as a result of an entire call processing. Then the CD agent is responsible for requesting the selected action to the MiTAI Gateway 25.
The System Management (SM) agent is responsible for managing other agents' status and sequencing call processing according to the system agents' priority. When the CD agent requests call processing, the SM agent scans the agents' status tuples in the TSpaces 37 and makes a sequence table according to their priority. It sends a processing request to the highest-priority-agent, waits for a response and then sends it to the next-highest-priority-agent. When the SM agent receives a response from the lowest-priority-agent, it sends an information tuple back to the CD agent.
The Context Simulation module serves as a context agent that dynamically detects, interprets and updates the user's current contexts. A Context Simulation window includes all of the possible contexts, which are set by a system administrator, and a user selects from them.
TSpaces 37 (i.e. tuple spaces), are implemented in one or more servers 23 as a network communication buffer with database capabilities. A more complete description of TSpaces 37 may be found in http://www.almaden.ibm.com/cs/TSpaces/, and in co-pending UK Patent Application No. 9920051.1. TSpaces 37 enable communication between applications and devices in a network of heterogeneous computers and operating systems. TSpaces 37 provides group communication services, database services, URL-based file transfer services, and event notification services. The TSpaces 37 are implemented in the Java programming language and thus automatically possess network ubiquity through platform independence, as well as a standard type of representation for all data types. The TSpaces system is appropriate for any application that has distribution or data storage requirements. It can perform many of the duties of a relational database system without imposing an overly restrictive (and primitive) type system, a rigid schema, a clumsy user interface or a severe runtime memory requirement. In the present invention, the TSpaces Server 23 is the medium between the system and user knowledge stores. However, it will be understood that Tspaces 37 may be replaced by a relational database or other shared database providing equivalent functionality for managing knowledge transactions, including read, write, update, take and scan, as well as event handling, such as registration and notification of events.
The MiTAI Gateway 25 facilitates communication to the MITEL telephony servers (i.e. PBX 26) for processes that are not based on the “C” development language. A description of the MiTAI Gateway 25 may be found in Ramiro Liscano, “MiTAI Gateway”, MITEL Networks, June 2001. The MiTAI Gateway 25 is a Windows based process that can be executed on any Windows platform. It can manage a single socket connection from any other process on a network and it supports a limited session protocol.
The MiTAI Gateway Server 25 is the intermediate system between the PBX 26 and the ACD's Application Level subsystem 35. The Application Level subsystem 35 registers an event to the MiTAI Gateway server 25 for the purpose of monitoring incoming calls.
In terms of system topology, the User Interface 31 is built on a Windows Platform, and it interacts with the Application Level 35 through an Event Handler 33. The Application Level 35 subsystems use TSpaces server 23 as the medium for communication and for accessing the System Knowledge Management of the server and clients.
All of the knowledge, which includes user information, user rules, the user's current context information, and call information, is stored in the TSpaces 37. The unit of storage is a tuple, as set forth in the references cited above, and as discussed in the Appendix to this specification.
User Information consists of basic user information, relationship information, a buddy list, user preference rules and the dynamically varying user's current context information. User Information is stored in a tuple named “UserProfile” and the structure is as follows:
There are two kinds of activities that can be defined. Some activities can be automatically detected by the system, whereas others can only be assumed or set by a user. For example, the system is able to know whether the user is ‘on the phone’, but it is difficult to judge if the user is ‘busy at work’ or ‘having a break’. Therefore, detectable activities are updated by the system automatically, and others are set by the user. A receiver's time context is set according to his time schedule. For example, if the user's lunchtime is scheduled from 12 p.m. to 1 p.m., the system can assume that the user is having lunch during that time period.
Call information is contained in a tuple that an agent shares to communicate with other agents for processing an incoming call. Therefore, it contains all of the necessary data fields for caller information and user preference rules. Agents take the “Call” tuple from the TSpaces 37 and update it according to their responsibility. For example, the RA agent assigns relationships between the caller and the receiver, the URA agent assigns all the appropriate user rules, and the UCR agent resolves user rule conflict by selecting only one user rule. The form of this tuple is:
It requests the TSpaces server 23 to notify the SM agent when a tuple is posted where the first field is “Call”, the second is “SMAgent”, and the fourth is user id, and where the third field is ‘new Field(String.class)’ which means that any value will be acceptable for this field.
Module interactions show the behavior of the classes, the modules, and the system as a whole. They describe how the components of the system interact by message passing, function calls, and by sharing the state information. The components interaction for the present invention are shown in the use case diagram and a state chart diagram of
In order to use the system, including the server system for administrators and the client system for users, a person must be authorized. First time users register by clicking the “Register” button in the “Log-in Window”, as indicated in
Registered users and administrators need to be authenticated before using the system. The fields for user ID and password in the “Log-in Window” have to be correctly filled in, and then the “OK” button clicked. If both fields are filled without omission, the system checks validation of each field. This validation procedure is the same as it for the user registration. The validated user-ID and password pair should be matched with those in the Tspaces 37. The system obtains the information by executing the “read( )” operation and comparing them. The login processes are finished when either the user clicks the “Exit” button or there is a match between the input user ID and password pair and the pair already in the Tspaces 37.
A prototype of the ACD system of the present invention has been implemented using the Java programming language on a Windows NT platform, with the following packages used for the implementation:
The details of the system's installation and execution method include unpacking the Java class files and executing them, as well as other server initiation processes which would be well known to a person of skill in the art.
The design of the ACD system is not restricted to any specific domain of users. A flexible method of defining knowledge for a user domain allows the system to be used in different domains. A system administrator can define a hierarchy of the user's location, activity and time according to a domain of target users. For the purposes of the successful prototype of this invention, the system provides two example domains: an office worker's domain and a professor's domain. A user may choose one of the two domains by clicking the corresponding button in the introduction window, as shown in
The ACD server system 27 is designed to be simple and easy to use. Having connected to the TSpaces Server 23, the installation procedure for the ACD server system 27 requires unpacking the Java class files and executing them on any machine on the network. At the start, a welcoming window provides brief information about the ACD system, administrator login information, and two buttons to load information for testing purposes: “An Office Worker Context Set Loading”, and “A Professor Context Set Loading”, as shown in
To login as an administrator, who controls knowledge and services for the server, the user is authenticated through the login window for an administrator as shown in
Once the login is authorized, the Server Main Window is presented for further processing, as shown in
Returning to
The server agents can be executed on any machine that has access to the TSpaces server 23. This means that any machine within the network can be used to execute a server agent. This design gives a flexible distribution of agents. All the agents can be executed together on a given machine by clicking the “ALL” button (
The Relationship Assigning (RA) agent assigns relationship information between the caller and the receiver based on the user's buddy-list. An example of execution is shown in
The User Rule Assigning (URA) agent assigns all user preference rules that match the conditions of the rules and the user's current context, as shown in
Rule Name: Worker-Office Rule
Rule Name: Thomas Ragan-Busy Rule
Rule Name: Worker-Office-Busy Rule
The names of the assigning rules are displayed as they are matched in
The User-rule Conflict Resolving (UCR) agent selects one user preference rule if there is more than one rule assigned by the URA agent. According to one embodiment of the invention, the UCR selects the most specific among the assigned rules. A rule with more conditions is considered more specific. In the scenario set forth above, the “worker-office-busy rule” is the most specific rule among the assigned rules, and is therefore selected as shown in the
As discussed above with reference to
During the login process (
A first-time user is prompted to provide basic user information such as user ID, password, name and phone numbers through registration. The input user ID is checked to see if it is not duplicated with existing user IDs. Each field has its own constraints of length and format. An error window notifies the user if any constraint is violated on clicking the “Register” button (
The main window for client control is presented if the login or registration process is successful, as shown in
The personal relationship information is shown as a tree structure that is easy to maintain (
Clicking any of the fields on the table in
As shown in
The user-rule Setting Window consists of three parts: the user rule table, which consists of sequence numbers and user rule names, UI buttons, and a Description window, as shown in
Clicking the “Add” button in the “User-rule Setting Window” starts the creation of a new rule. Adding a new user rule involves four steps. The first step is to select the contexts as a part of the conditions of the rule being created (
The second step is selecting a type of caller as a part of the condition. One of three categories can be selected: any caller, a buddy list table, and a relationship tree. These three categories are exclusive, so that radio buttons are provided to select only one category. When a category is selected, the user can then select items in its selection window.
The third step is selecting an action of the rule from the predefined list of actions, as shown in
The fourth and the last step for creating a new rule is confirmation. As shown in
The description of the rule as an example for creating a new rule (from FIGS. 23 to 26) is as follows:
Rule Name: Mindy Baker-Office Room-Busy Rule
The entire process to create a user preference rule is shown in
Ultimately, the user's current contexts such as the current location and activity are updated by the Context Agent. In the successful prototype, a simulation program was used to substitute the occurrence of events in real-life. For testing purposes, the tester selects one of the desired contexts on the hierarchy trees, and then clicks the “Apply” (
As discussed above, the client has two agents: the Call Delivery Agent (CD) agent and the System Management Agent (SM) agent. Each agent has its own display window to present procedural messages to a user. The CD agent is connected to both the TSpaces server 23 to communicate with other agents and to the MiTAI Gateway server 25 to communicate with the phone system.
The window in
A call processing request can be received from either the Call Simulation or the Call Monitor. The Call Monitor communicates with the MiTAI Gateway server 25 for handling actual phone calls whereas the Call Simulation is another window on a server machine to test the system without MiTAI Gateway interfaces. When call processing has finished involving all the available agents, the CD agent extracts the selected user rule, which is a result of the processing, and requests the Call Monitor to carry out the action stated in the selected rule. When the example in
The SM agent is also connected to the TSpaces server 23 to communicate with other agents. The display in
In summary, according to the present invention a model of contexts is defined for a message delivery system and a system architecture provided that facilitates the creation of context- and rule-based communication. Location context is used to forward incoming calls based on location information. The user's activity or availability is used in the instant invention to notify other connected users about the user's status such as “busy”, “be back”, “away” and “lunch”. The time context is used to set the time of application of certain user-defined rules.
By receiving the appropriate messages in the right situation, system users benefit from minimal interruptions. By exploiting personal traits and filtering messages based on both a user's current context models and his/her preference rules, the possibility of having desirable delivery action for the user is enhanced. Although the user's current contexts have been simulated for the purpose of the working prototype of the system set forth in detail above, a person of skill in the art will readily appreciate that the system can be implemented with a Context Agent, which actually detects a user's contexts. To that end, a simple type of the Context Agent, which detects a computer's mouse movement, has been tested. In operation, a user who uses multiple machines in the network first logs into a particular computer. The Context Agent detects the computer's mouse movement and in response updates the user's location information in the Tspaces 37 so that an incoming call can be notified or forwarded to the user at that location.
As discussed in our co-pending application filed on the same day as this application and entitled “Privacy and Security for Presence Systems with Tuple Spaces”, although using the TSpaces 37 provides great flexibility in the aspect of multiple agent system design, it suffers from a weakness in security since it allows for the sharing of all information. Some of the privacy-sensitive information such as the user profile should be protected. The TSpaces server 23 provides access control by setting user and group permissions on the Tspaces 37 so that only users with the proper access control permissions may read and write tuples from the TSpaces. Additional security measures are set forth in the aforenoted copending application.
Also, although the successful prototype has been described in terms of call processing, it is contemplated that the principles of the invention may be extended to implement context-based event handling beyond telephony, such as email processing, visitor notification services, etc.
All such modifications and variations are believed to be within the sphere and scope of the present invention as defined by the claims appended hereto.
The following Appendix sets forth in greater detail the structure and operation of specific tuples used to implement the successful prototype:
Appendix: Structure of Knowledge
Knowledge is stored in the Tspaces 37. The unit of knowledge is a “tuple”, which is also the unit of data transaction with the TSpaces. A tuple is an aggregation of field(s), and maximum of seven fields are allowed by the TSpaces. The first field is reserved for the name of a tuple, and other fields can be either a simple string or a string with tags1. The simple string will be marked as ‘String’, and the string with tags will be marked as ‘Tag String’ in the definitions below. There are four tuples for this system: ‘UserProfile’, ‘Call’, ‘Context’, and ‘Agentinfo’.
1Each field contains simple, but flexible data. It uses tags similar to XML format, but it does not use any particular parser due to the problems of TSpaces with XML. It uses internal function to have access to data instead of parsing.
There are user-defined fields: the ‘relationship ’ in the ‘UserProfile’, the ‘location’ and the ‘activity’ in the ‘Context’. The knowledge structure for those fields will be defined during run-time by using the Guls. Those fields will be marked as the ‘User Defined’. Some fields allow saving many same-format-fields. For example, the ‘userrule’ field in the ‘UserProfile’ tuple allows saving many user rules in that field. (marked as *)
1. User Profile
1.1 Description
This is user specific information, includes basic user information, for call delivery.
1.2 Tuple Fields
userinfo Field: This field contains fundamental user information for operating the system.
buddylist Field: This field contains a list of buddy and their information, which is set by a user
userrule Field: This field contains a list of user rules, which are created by a user
context field: This field contains the user's current contexts, which are set by either simulation agent or context agent.
This tuple is for call processing among agents.
2.2 Tuple Fields
callinfo Field: This field contains basic information of both a caller and a receiver such as phone number and name. It also contains relationship information between them.
This is the hierarchy of context information that will be decided and input by system administrator. The contexts, which will be used in our prototype system, are a user's location, his/her current activity and the present time.
3.2 Tuple Fields
4. AgentInfo
4.1 Description
Agents are responsible for updating their status into the TSpaces. The SM agent uses this agent information when it decides the sequence of the process according to their availability.
4.2 Tuple Fields
| Number | Date | Country | Kind |
|---|---|---|---|
| 0218712.8 | Aug 2002 | GB | national |
