Mail Room Automation

Expert system for the automatic management of certified electronic e-mail

The new recent regulations on certified electronic e-mail impose a series of obligations on the public administration that could send the current organizations into crisis. Certified electronic e-mail may become something that complicates rather than simplifies the original idea of its inventors. Through its technology Idoq is capable of relieving the public administration from the manual management (receipt, sorting, first response, categorization and storage, etc.) of certified electronic e-mail. BRAINCAT, an automatic categorizer, is at the base of the solution, a tool that allows for the autonomous association of a free or formatted text to one or more of a set number of categories. Many possible variants exist, and it is only in the analysis of the context in which the categorizer and its objectives - that allow for implementation of the most appropriate technology - are applied. The categorizer is broadly configurable for adapting itself in the best possible way to the greatest variety of contexts. Therefore, just to give an example, here is a list some of the system's functions.

• Automatic routing for texts with content relevant for the purposes of turning on the alert: in this case, the system is configured with a series of phrases associated to concepts that characterize the documents that must be "captured" by the system. For example, if the goal were to have the e-mails that complain of delays or anomalies for a certain service stand out, sentences associated to this concept are collected. Through the advanced logics of contextualization and of semantic equivalences, the categorizer will be able to execute an abstraction of these sentences to capture the most texts possible that refer to this type of reports.
• Recognition of the type of text based on content: if one must divide up the texts based on the language in which they are written or the argument they deal with (for example, "disability pensions" or "severance indemnity") and so on, we will use a strategy based on the collection of a sufficiently high number of example texts and on the training of the system. Self-learning and statistic criteria strategies are adopted through decomposition and mapping of the entire text in its base elements (words, lexical roots, etc.).
• Recognition of the type of text based on structure: if the content does not clearly distinguish the type and the parts of the text, it is necessary to use the layout and the structure of the text itself. In this case, example document sets and implementations based on self-learning and statistic criteria by spatial mapping of a series of key words, semantic annotations or other identifiable reference elements in the text are employed.
• Automatic definition of categories: the system also allows for the activation of the automatic process of category creation directly through reading and interpretation of the documents. In fact, if the objective is to automatically create groupings in which to insert similar texts, clustering techniques and self-organizational maps to automatically create the categories will be used rather than defining them manually.

The management HUB of the certified electronic e-mail is capable of:

• Managing the arrival of the certified electronic e-mail by going and identifying and marking the sender and all necessary information necessary to fully tracing the event. At this point in the application's process, it will be able to carry out a series of congruity, legitimacy, and consistency checks of the elements identified in the certified electronic e-mail.
• Activating a multichannel integrated system of reporting and information on the event.
• Acquiring and structuring the contents of the communication in the proper fields, such as the subject of the message, the text of the communication, any attachments treated differently based on their nature (text file, TIF, PDF, XMLS, structured files, etc.)
• Executing all the necessary operations to making the contents of the communication homogeneous using different OCR instruments (typewritten text, manuscript, bar codes, images, etc.)
• Activating engines for the recognition and comprehension of the contents of the communication. In this stage, our highly valuable solution employs two technologies of reference: neural engines and semantic categorizers. The outcome of this stage is the automatic identification of the communication's content and its categorization inside the areas of competence that the receiving structure divides itself into.
• Send the communication to the appropriate manual categorization station in the event in which the automatic categorization wasn't able to reach the established levels of reliability. The percentage of communications to be enriched manually can vary greatly, but our experience in semi-structured situations (institutions with specialized competences), is of an automatic sorting greater than 60% with peaks even at 80%. Downstream of the categorization, the work flow system sends the communication to the competent Organizational Unit. In this stage, mechanisms of direct integration with the individual public administrations' document and transactional systems can be created, and multichannel alert mechanisms can be activated. Our system is based on a SOA platform (Enterprise Communication Suite) capable of dialoguing on various levels with external systems and of orchestrating interoperability with more than one system at the same time.
• All of the documents acquired will be automatically sent to the electronic storage besides being inserted in the public administration's document systems. Should the public administration not have its own document system, our application offers an internal archiver for the management of documents and automatically proceeds with their indexing.
• The solution proposed faces the issue of integration with the public administration's transactional and document programs through a broadly customizable and programmable middleware system. The automatic export system acts on all of the internal and external communications and has a special built-in adapter for the most common document and transactional systems.
• Through a System Monitor dedicated to the application, it is able to trace all of the events and report any anomalies of each process in real time. The system monitor is also equipped with multichannel automatic alert functions.
• A Performance Monitor completes the system, allowing automatic management of reports and management of Performance Indicators for each stage of the process. The Analytics module continuously monitors the crossover time of the documents and the users' activities, pointing out delays in document treatment or process anomalies.