Patent Application
20190266254
An enterprise may need to respond to requests. For example, an insurance company might need to respond to customer requests for Certificates Of Insurance (“COI”), policy cancelations, etc. Note that these requests might be received via various channels, such as a telephone call, web form, email message, etc. Responding to these requests can be a time consuming and difficult task, especially when a substantial number of requests are received (e.g., an enterprise might receive thousands of requests on a daily basis). Moreover, different types of requests might require different types of processing (e.g., information validation, determining supplemental information, etc.). Determining how to properly respond to requests can also be a time consuming and error-prone process, especially there are many different types of requests that might need responses.
Thus, there is a need in the art for methods and systems to respond to requests in an efficient and accurate manner.
According to some embodiments, systems, methods, apparatus, computer program code and means are provided to respond to requests in an efficient and accurate manner. According to some embodiments, an artificial intelligence orchestration platform may automatically determine an intent of an electronic record (e.g., an email message, text, translated voice channel request, etc.) associated with a request (e.g., by communicating with a classification platform service or analyzing the electronic record). Based on an indication of intent, an entity extraction platform may extract at least one requisite entity identifier from the electronic record in accordance with a transaction requirement. A robotic automation platform may then process the request utilizing the indication of intent and the extracted requisite entity identifier. For example, the robotic automation platform may transmit a complete response to the request, pre-populate data in a template provided to a human knowledge worker, determine additional information associated with the request, etc.
Some embodiments provide: means for automatically determining, by an artificial intelligence orchestration platform, an intent of an electronic record associated with a request; based on an indication of intent, means for extracting, by an entity extraction platform, at least one requisite entity identifier from the electronic record in accordance with a transaction requirement; and means for processing the request, by a robotic automation platform, utilizing the indication of intent and the extracted requisite entity identifier.
A technical effect of some embodiments of the invention is an improved and computerized way of responding to requests in an efficient and accurate manner. With these and other advantages and features that will become hereinafter apparent, a more complete understanding of the nature of the invention can be obtained by referring to the following detailed description and to the drawings appended hereto.
The present invention provides significant technical improvements to facilitate electronic messaging and dynamic data processing. The present invention is directed to more than merely a computer implementation of a routine or conventional activity previously known in the industry as it significantly advances the technical efficiency, access and/or accuracy of communications between devices by implementing a specific new method and system as defined herein. The present invention is a specific advancement in the area of electronic record analysis by providing benefits in data accuracy, data availability, and data integrity and such advances are not merely a longstanding commercial practice. The present invention provides improvement beyond a mere generic computer implementation as it involves the processing and conversion of significant amounts of data in a new beneficial manner as well as the interaction of a variety of specialized client and/or third-party systems, networks, and subsystems. For example, in the present invention information may be transmitted to remote devices from an orchestration server and electronic records may be interpreted and routed as appropriate, thus improving the overall performance of the system associated with message storage requirements and/or bandwidth considerations (e.g., by reducing the number of messages that need to be transmitted via a network). Moreover, embodiments associated with automatic predictions might further improve communication network performance, user interactions, real time chat or telephone call center responsiveness (e.g., by better preparing and/or allocating resources), etc.
In other cases, an email message 120 might be reviewed by a human indexer 122 who will attempt to determine what the customer needs. The indexer can then manually create a work order 124 that is then manually processed 126 before the request is completed 190. Such an approach might, for example, cost $5.00 per-request to process. In still other cases, a phone call 130 might be received by a human customer service representative 132 who will manually process the request 134 to completion 190. Such an approach might, for example, cost $10.00 per-request to process.
Note that an enterprise may receive email, text, speech, etc. requests from customers for various reasons. For example, an insurance company might receive emails from customers regarding the purchase of new insurance policies, billing questions, inquiries about insurance claims, etc. Moreover, different service representatives may have different skills, abilities, domain expertise, etc. For example, one service representative might specialize in helping customers purchase new insurance policies while another service representative specializes in helping answering customer billing questions. Thus, a received email may need to be eventually routed to an appropriate queue to be serviced by a customer service representative.
In the approach of
The indexer might use one or more displays to process the request. For example,
Using such a display 300, however, can be a time consuming and error-prone process (e.g., an incorrect transaction type 320 might be selected). According to some embodiments described herein, systems, methods, apparatus, computer program code and means may provide a tool to facilitate the full end-to-end automated processing of email/text/voice channel requests. In some embodiments, an email is received. An orchestrator/flow manager may be accessed, from an artificial intelligence platform, to identify customer intent. It may then be arranged for requisite entities, identified by intent type, to be extracted in accordance with transaction requirements. Once the requisite data is acquired, a robotic automation may complete the request or pass along information to knowledge workers with portions of the data administration complete. In some cases, the robotic automation may reach out to customer, or other accessible backend systems, in an attempt to acquire missing information or to correct invalid data (e.g., a typographical error in a policy number).
Note that the robot assessment 414 might require a structured intake of data. According to some embodiments described herein, Natural Language Process (“NLP”) may remove the need for human indexers and create structured intake data from free-form email requests. In particular, a Natural Language Classifier (“NLC”) service 442 may receive email, text message, translated voice input, etc. A Natural Language Understanding (“NLU”) text extraction process 444 may then locate the information needed by the robotic assessment 414 to complete the request 490.
The NLC 442 or NLU 444 might be, for example, associated with a cloud-based architecture, Personal Computer (“PC”), laptop computer, an enterprise server, a server farm, and/or a database or similar storage devices. The NLC 442 or NLU 444 may, according to some embodiments, be associated with a business organization or an insurance provider.
As used herein, devices, including those associated with the NLC 442 or NLU 444 and any other device described herein, may exchange information via any communication network which may be one or more of a telephone network, a Local Area Network (“LAN”), a Metropolitan Area Network (“MAN”), a Wide Area Network (“WAN”), a proprietary network, a Public Switched Telephone Network (“PSTN”), a Wireless Application Protocol (“WAP”) network, a Bluetooth network, a wireless LAN network, and/or an Internet Protocol (“IP”) network such as the Internet, an intranet, or an extranet. Note that any devices described herein may communicate via one or more such communication networks.
According to some embodiments, an “automated” NLC 442 or NLU 444 may mine information from a request (e.g., text input data sources). As used herein, the term “automated” may refer to, for example, actions that can be performed with little or no human intervention. The NLC 442 or NLU 444 may store information into and/or retrieve information from databases. The databases may be a locally stored relational database or reside remote from the NLC 442 or NLU 444. The term “relational” may refer to, for example, a collection of data items organized as a set of formally described tables from which data can be accessed. Moreover, a Relational Database Management System (“RDBMS”) may be used in connection with any of the database tables described herein. According to some embodiments, a graphical administrator interface may provide an ability to access and/or modify the NLC 442 or NLU 444. The administrator interface might, for example, let an administrator define terms, dictionaries, mapping rules, etc. associated with text mining. Moreover, note that the NLC 442 or NLU 444 may operate asynchronously and/or independently of other insurance applications.
Although a single NLC 442 or NLU 444 is shown in
In this way, the system 400 may facilitate an accurate and efficient response to customer requests. For example,
At 510, the system may receive an electronic record associated with a request 510. The electronic record might be associated with, for example, an email message, a text message, a voice channel request that has been translated into text, a fax, a video request, a web site submission, a mobile application, a messaging application, etc.
At 520, an artificial intelligence orchestration platform may automatically determine an intent of the request associated with the electronic record. For example, the artificial intelligence platform might communicate with a classification platform service and/or analyze the electronic record to determine the intent. Based on an indication of intent, at 530 an entity extraction platform an entity extraction platform may extract at least one requisite entity identifier from the electronic record in accordance with a transaction requirement. According to some embodiments, the entity extraction platform is coupled to a domain lexicon specific to intent type. Moreover, the entity extraction platform may access e-forms, scanned forms, online web forms, etc. for data extraction on intent type element extractions.
At 540, a robotic automation platform may process the request utilizing the indication of intent and the extracted requisite entity identifier. For example, the robotic automation platform may process the request by automatically transmitting a complete response to the request. In other cases, the robotic automation platform may process the request by automatically pre-populating data in a template provided to a human knowledge worker. In still other cases, the robotic automation platform may process the request by automatically determining additional information associated with the request (e.g., by asking an originator of the request for the additional information or receiving the additional information from a third-party device). Note that the robotic automation platform may interacts with downstream systems through Application Programming Interface (“API”) and/or front end Robotic Processing Automation (“RPA”) to transact customer requests.
According to some embodiments, the system may further include an intent library storing, for each of a plurality of customers of an enterprise, historic customer request and intent definition information. Moreover, embodiments might include an artificial intelligence customer service terminal to facilitate interactions with a customer and/or a parsing platform to decompose large Binary Large Object (“blob”) text into smaller sentence structures for finite intent understanding and handling multiple intent requests within a single blob of text.
According to some embodiments, a supervised learning platform may be provided for low confidence transactions to facilitate model training and ongoing systematic model enhancements. Similarly, an unsupervised learning platform might be provided for low confidence transactions to facilitate model training and ongoing systematic model enhancements to systematically leverage knowledge worker request processing. Moreover, an analytics platform might identify type I and type II errors in classification, maintain statistics on historic decisions, and/or generate final decision processing logs.
Thus, embodiments may provide a “straight through” processing engine that is able to take in different types of transactions, determine what data to extract and how to extract it, determine how to route the information, extract appropriate data, and route a structured work order to a robot for automatic processing. Moreover, embodiments may provide an ability to plug and play “data extraction” technologies including: (i) an engine with workflow capabilities that knows what to do with each transaction, and (ii) an engine that passes classified transactions, with data extracted, to a robot to complete the request. Use might include, for example: email service requests (such as request for a COI), (ii) processing of claims with photographs of damage, and (iii) processing of claims with video of damage. The architecture described herein may enable addition of whatever technology is needed to do data extraction and classification. Moreover, embodiments may enable any combination of these technologies to be applied to a transaction to achieve the needed level of data extraction and classification.
The embodiments described herein may be implemented using any number of different hardware configurations. For example,
The processor 1410 also communicates with a storage device 1430. The storage device 1430 may comprise any appropriate information storage device, including combinations of magnetic storage devices (e.g., a hard disk drive), optical storage devices, mobile telephones, and/or semiconductor memory devices. The storage device 1430 stores a program 1412 and/or an orchestration engine 1414 for controlling the processor 1410. The processor 1410 performs instructions of the programs 1412, 1414, and thereby operates in accordance with any of the embodiments described herein. For example, the processor 1410 may automatically determine an intent of an electronic record (e.g., an email message, text, translated voice channel request, etc.) associated with a request (e.g., by communicating with a classification platform service or analyzing the electronic record). Based on an indication of intent, the processor 1410 may extract at least one requisite entity identifier from the electronic record in accordance with a transaction requirement. The processor 1410 may then arrange for a robotic automation platform to process the request utilizing the indication of intent and the extracted requisite entity identifier. For example, the robotic automation platform may transmit a complete response to the request, pre-populate data in a template provided to a human knowledge worker, determine additional information associated with the request, etc.
The programs 1412, 1414 may be stored in a compressed, uncompiled and/or encrypted format. The programs 1412, 1414 may furthermore include other program elements, such as an operating system, a database management system, and/or device drivers used by the processor 1410 to interface with peripheral devices.
As used herein, information may be “received” by or “transmitted” to, for example: (i) the apparatus 1400 from another device; or (ii) a software application or module within the text mining apparatus 1400 from another software application, module, or any other source.
In some embodiments (such as shown in
Referring to
The request identifier 1502 may be, for example, a unique alphanumeric code identifying a request received from a customer (e.g., an email, text message, facsimile, etc.). The primary intent 1504 might be automatically generated using NLC and/or NLP technologies. The entities 1506 might be one or more identifiers (e.g., names, addresses, etc.) that were automatically extracted from text associated with the request identifier 1502.
A work routing process routes the request to an in-basket at 1640. If the request is associated with a cancelation at 1642, a robotic automation completes the cancelation at 1650. If the request is associated with a COI at 1644, a robotic automation sends the COI to an appropriate entity at 1652. If the request was associated with a cancelation or COI, it may be routed to a human representative for manual processing.
Note that a request might include one or more attachments (e.g., files) in addition to email text.
Thus, embodiments may provide a system and method for email and text processing using artificial intelligence and domain lexicon. Moreover, embodiments might read emails (including SMS text and transcribed speech), interpret intents, extract element content, and/or fully process customer transactions with limited human intervention at run-time. Note that embodiments may leverage NLP, container and application orchestration, and RPA tools to integrate disparate technologies as a suite in pursuit of automating unstructured requests through various channels. Some embodiments may utilize third party data science and orchestration services. Moreover, model accuracy may be assessed using blind production emails within a test suite.
The following illustrates various additional embodiments of the invention. These do not constitute a definition of all possible embodiments, and those skilled in the art will understand that the present invention is applicable to many other embodiments. Further, although the following embodiments are briefly described for clarity, those skilled in the art will understand how to make any changes, if necessary, to the above-described apparatus and methods to accommodate these and other embodiments and applications.
Although specific hardware and data configurations have been described herein, note that any number of other configurations may be provided in accordance with embodiments of the present invention (e.g., some of the information associated with the databases described herein may be combined or stored in external systems).
Note that information might be provided to customers and/or administrators in any number of different ways. For example,
Applicants have discovered that embodiments described herein may be particularly useful in connection with particular types of insurance policies and associated claims. Note, however, that other types of business and insurance data may also benefit from the invention. For example, embodiments of the present invention may be used in connection with automobile insurance policies, financial services, government functions, etc.
Moreover, it is also contemplated that embodiments may process recommendations in one or more languages, such English, French, Arabic, Spanish, Chinese, German, Japanese and the like. In an exemplary embodiment, a system can be employed for sophisticated text analyses, wherein text can be recognized irrespective of the text language. The relationships between the various words/phrases can be clarified by using a rules engine for classifying words/phrases as a predictor of intent or entity.
According to some embodiments, text data may be used in conjunction with one or more predictive models to take into account a large number of intent, entity, and/or other parameters. The predictive model(s), in various implementation, may include one or more of neural networks, Bayesian networks (such as Hidden Markov models), expert systems, decision trees, collections of decision trees, support vector machines, or other systems known in the art for addressing problems with large numbers of variables. Preferably, the predictive model(s) are trained on prior text data and outcomes known to the insurance company. The specific text data and outcomes analyzed may vary depending on the desired functionality of the particular predictive model. The particular text data parameters selected for analysis in the training process may be determined by using regression analysis and/or other statistical techniques known in the art for identifying relevant variables and associated weighting factors in multivariable systems. The parameters can be selected from any of the structured data parameters stored in the present system, whether the parameters were input into the system originally in a structured format or whether they were extracted from previously unstructured text, such as from big data.
Some embodiments have been described herein with respect to COI and cancelation request. Note, however, that any embodiments might also be applicable to, for example, Automobile Liability (“AL”) and/or General Liability (“GL”) insurance scenarios. For example, a lack of structured data may present challenges (such as when medical bills are not paid directly by an insurance enterprise or demand packages take years from an original Date Of Loss (“DOL”)).
The present invention has been described in terms of several embodiments solely for the purpose of illustration. Persons skilled in the art will recognize from this description that the invention is not limited to the embodiments described, but may be practiced with modifications and alterations limited only by the spirit and scope of the appended claims.
