Conference heron::euro_swas_ai

324.0. "YES" by ULYSSE::ROACH (TANSTAAFL !) Fri Apr 19 1991 12:37

 

                  I N T E R O F F I C E   M E M O R A N D U M

                                        Date:     12-Apr-1991 07:25pm CET
                                        From:     Ed Killeen
                                                  KILLEEN.ED AT A1 at BUFFER at OGO
                                        Dept:     New Ventures
                                        Tel No:   276-8376

TO: See Below

Subject: APPOINTMENT OF NEW INTERNATIONAL BUSINESS MANAGER FOR IIS


STOW, MA

    In the last few years Digital's New Ventures Group has been engaged in the     
    development of new technologies related to intelligent information systems.     
    Although the Group does not develop products, it was successful in 
    developing several service offerings that address many customer needs 
    and which, when sold, have also leveraged many other Digital products both 
    hardware and software.

    These services include, but are not limited to, intelligent information
    systems, content-based retrieval systems, image-based document retrieval 
    systems, electronic drawings management systems, indexing and retrieval 
    of news stories, image character recognition, automated forms processing 
    systems, manufacturing process control solutions and more. 

    These services utilize state-of-the-art technologies such as neural 
    networks, image processing, pattern recognition, artificial 
    intelligence and others, developed by Digital through third-party 
    alliances and through participation in R&D consortiums.

    The Group is also working with the Database Systems Research Group 
    in Colorado Springs to incorporate DIDDLY (a multi-media offering and 
    one of the most successful demonstrations at DECWORLD-90) in customer 
    applications.

    The above business has been extremely successful for the New Ventures 
    Organization and has grown several folds in the two years of its 
    operation. It has also incremented over $10 million dollars in hardware 
    and software sales for the Corporation.

    Recently, New Ventures decided to share its technologies and to 
    replicate its successful models with any international organization that 
    wishes to benefit from our experiences and successes.

    Two international EICs that have already sent their engineers for 
    training in the U.S. are from the AI IST in Valbonne, France, and the 
    ACT in Munich, Germany. These centers have already started to implement 
    such solutions with their customers. The Group has also conducted 
    technology training in Tokyo, Japan and Valbonne.

    To facilitate technology transfer more expeditiously and effectively, 
    the Intelligent Information Services Group (IIS) has appointed Matti 
    Jaffe to the role of International Business Development Manager, 
    effective immediately.

    Matti has worked at New Ventures since its inception and most recently 
    was the manager of Automated Forms Processing Services and Machine 
    Learning Consulting Services.

    In his new role Matti will be responsible for the successful
    implementation of IIS technologies and services abroad, developing
    operational models for successful transfer of Digital's new ventures
    technologies across international borders, and adoption of intelligent
    technologies developed by international third-parties across
    international borders.

    Matti will be working with the various GIA and European sales and sales 
    support organizations to understand their customers' "intelligent 
    information services" requirements and support them in their search for 
    solutions.

    The following pages describe a sample portfolio of services that IIS has 
    performed to date. If you have any questions about support, logistics or
    details about programs described, please do not hesitate to give Matti 
    a call at DTN: 276-8931, or drop him a note at BUFFER::JAFFE or MATTI
    JAFFE @OGO.

    Please join us in wishing Matti success in this important new role.





                        INTELLIGENT INFORMATION SERVICES

                             PORTFOLIO OF SERVICES

APPLICATION:    ON-LINE INFORMATION COLLECTION, DISTRIBUTION, CONTROL,
                REVIEW AND PUBLISHING SERVICES FOR PRIVATE INVESTORS

PROBLEM: 

         Large information providing companies are faced with multi-feeds
         of information and data (wire services, press releases, stock 
         reports, etc.) from both internal and external sources. 

         Present information gathering is either manual or semi-automated 
         and is labor intensive and complex. Gathering knowledge involves 
         steps of collecting, reviewing, summarizing and editing of
         information that will become available to customers.

APPROACH:

         o Graphical user interface incorporating control functionalities
           with variants of interfaces for editorial and publishing
           functions

         o Central collection of all feeds with appropriate conversion to 
           electronic formats

         o Automated data extraction, indexing and database look-up

         o Incorporates text and image

ADVANTAGES:

         Gaining control over in-coming information and having it
         centralized enables the customer to cut costs, increase
         throughput, have immediate access to information, and conceive 
         new product offering.

STATUS:  Project is in process. Phase-1 (information collection and review)
         is scheduled to be completed in June, 91, Phase-2 (information 
         editing and publication) is scheduled to be completed in Sept., 91.




APPLICATION:   ENGINEERING DRAWING/DOCUMENT MANAGEMENT SYSTEM (EDMS)
               FOR OIL REFINING OPERATIONS

PROBLEM: 

         Refining maintenance organizations require multitude of
         documents, procedures, manuals, technical specifications -- all 
         of which are derived from various engineering organizations and
         which are closely interrelated. 

         This environment has created a need to integrate documents and
         procedures (typically A-size documents) with existing CAD
         engineering documents as well as scanned images of E-size
         drawings, and dynamically maintain relationships among them,
         while facilitating retrieval, review, feedback, and updating of
         all affected documents for a variety of purposes. 

APPROACH:

         Integrate two existing third-party products (Excalibur's PixTex 
         and ABB's Re: Solution) under a common graphical user interface 
         to address this broad requirement.

ADVANTAGE:

         Fast access to engineering and maintenance information based on 
         both key words and fuzzy search.

STATUS:  In advance stages. Test-site implementation -- completed. Pilot
         implementation -- Q4/FY91, production rollout -- FY92.




APPLICATION:   COMPUTER-INTEGRATED NEW DRUG APPLICATION (CINDA)

STATEMENT:

        New Drug Application (NDA) is a long and complex procedure which
        involves multiple disciplines where the integrity of the entire 
        process is reflected in the completeness, quality, and
        availability of documents created from the point that basic 
        research starts to the point where clinical trial data is compiled 
        and published as the actual NDA. Keeping documents in a structured 
        repository can accelerate the NDA process.

APPROACH:

        Working through The Pharmaceutical Industry Advisory Board,
        embarking on an extensive proof-of-concept to identify Digital
        and third-party hardware and software products which can be
        integrated to meet the needs across the entire NDA spectrum.

        The application addresses various databases incorporating text
        and scanned images of documents and publications. 

BENEFITS:

        The justification of such application is the accelerated return
        on investment for a NDA, thus increasing the profitability window
        for the drug company.

STATUS: Project started. Proof-of-concept be completed in Q4/FY91




APPLICATION:        ON-LINE DOCUMENT MANAGEMENT FOR LOAN PROCESSING 

PROBLEM:

        The organization processes millions of paper documents annually, 
        presently done manually. The company also processes hundreds of 
        daily requests for information from files for universities and
        students and cannot keep up with demand with manual procedures.

APPROACH:

        Provide a distributed on-line electronic filing solution which
        involves scanning, OCR, and content-base retrieval.

ADVANTAGES:

        An efficient system that is more responsive and reliable with
        access to relevant documents (some of them unknown at the outset!).
        Ability to perform better searches as compared to searches based on 
        Boolean clues.

STATUS: Project started. Demo to be completed by Q1/FY92




APPLICATION: IMAGE-BASED CHARACTER RECOGNITION OF CREDIT CARD APPLICATIONS

PROBLEM: 
        
        The customer processes 6,000 credit card applications daily,
        employing many costly key-entry operators. Information on the
        applications is hand-written. Due to marketing requirements the
        information zones cannot be blocked out with half-tone ink.

APPROACH:

        Utilize standard OCR to recognize machine-printed characters
        while utilizing a sophisticated Machine Learning-based algorithm 
        to recognize free-style hand-written numeral characters.

        Image-based recognition results are correlated against database
        information for added accuracy.

ADVANTAGES:

        o Reduced operational costs

        o A modular approach that can grow with customer's added volume

        o Image-base solution improves productivity, data integrity and 
          services, and adds flexibility in storage and retrieval 
          requirements

STATUS: Project started. To be completed by June, 91.




APPLICATION:       CLAIMS PROCESSING WORKSTATION

PROBLEM:

        Billions of medical related forms are processed each year at a
        major cost to the organizations who process them manually.
        In the process claim forms are lost and missfiled, payments are 
        duplicated, and claims' adjudication could be faulty.

        This problem is mirrored across many other industries.

APPROACH:

        Team up with a third-party that has developed software for forms
        processing and integrate its products to a broad solution which
        would incorporate various scanner interfaces, form management, OCR, 
        workflow, and a storage manager.

ADVANTAGES:

        o High-performance, reliable and feature-loaded system

        o Coverage of a variety of scanners

        o Drastically reduces operational costs and improves productivity
          and service

        o Modular NAS system based on various servers

STATUS: Started. To be completed by 2/92.




APPLICATION:    SENSOR FUSION OF A LARGE METAL PROCESSING OPERATION

PROBLEM: 

         In a large discrete or continuous processing operation sensors are
         utilized to measure and control processes; in most cases this is 
         done under real-time conditions. 

         The operation has been collecting data for years but has lacked 
         the knowledge of how to process it intelligently. One of the 
         reasons is that many sensor data interrelationships are non-linear 
         and difficult to model.
         
         Another problem is that a complex operation needs to address 
         hundreds (and sometimes thousands! of sensors), all interrelated 
         and in real time.

APPROACH:   

         In its approach, New Ventures has used various Machine Learning 
         techniques to filter useful information from the real-time data 
         relationships between process states and process parameters. This 
         information is then used as feedback to optimize and improve the 
         process.

         For systems that are closely coupled (already integrated), our
         approach is integrated as a decision support and becomes part of 
         the total system. The decision support passively extracts data on 
         on-going processes, performs analysis, and through the latest 
         graphical and windowing techniques aids the operator in his complex 
         tasks of controlling the whole operation.

ADVANTAGES:

         o Solution is cost-effective and very easy to justify in any 
           operation that has collected historical data from process sensors.

         o Seamless solution that can address very complex problems with
           relatively minimal initial investment. Once installed the 
           solution does not disturb daily on-going operations.

         o Solution is based on adaptive modeling which is more accurate 
           than direct process modeling of sensor interrelationship and 
           yields better performance.

STATUS:  Finished Phase-1 successfully -- concept demo. Moving into 
         Phase-2 -- Final implementation, to be completed by 7/91.
         



APPLICATION:   THEASAURUS-BASED INTELLIGENT INFORMATION RETRIEVAL SYSTEM

PROBLEM: 
        Information might reside in multiple files and in distributed
        heterogeneous databases, requiring a common user-interface. Content 
        for existing and future databases needs to be defined in order to 
        choose appropriate information sources and types.


APPROACH:

        A customer problem analysis including content and technical feasibi-
        lity is performed. A Thesaurus Management Architecture acts as a 
        knowledge-base to describe system content through term definition 
        and relationships. 

        The Thesaurus Application Programmers Interface (API) integrates
        existing databases and supports new applications as they are built.

ADVANTAGE:

        The Thesaurus Management Architecture is the common index which 
        supports efficient maintenance and retrieval of multiple databases. 
        Resident expertise supports on-line production systems which maintain 
        and distribute business intelligence. 

STATUS: The process is in place to define information and technical needs.
        Phase-1 (Q4/FY91), includes a conceptual plan outlining information
        sources, types, and architecture required to support the application.
        Phase-2 (Q1/FY92) includes system design and prototype implementation.

Distribution:


TO:
DAVE BACKMAN @AKO   DIRK BADE @UFC      MIKE BRIDGES @BST   CLIFF CHATTEN @RDL  
PETER COOK @REO     ALBERTO COSTALES @A BRUNO D'AVANZO @GEO VINCENT D'SOUZA @GE 
HARI DAS @AKO       HANS DIRKMANN @GEO  PIER CARLO FALOTTI  RONALDO FORESTI @AK 
ERIC GERELLE @GVO   SERGIO GIACOLETTO @ LYNNE GILLON @MCL   JOHN GOUGH @GMO     
JOHN GREEN @SNA     LUIS GUTIERREZ @FLO CARL HEMINGWAY @VBE LENNART HJORD @SOO  
ROBERT HO @MEL      YUTAKA IIDA @TKO    YOSHINORI ISHII @TK WOLFGANG JAEGER @SQ 
STEVE JONES @VBE    SHINGERUY KAN @TKO  NAOKI KAWAMOTO @OSA KEES JAN LINDEMAN @ 
GARRY LOWE @AKO     ED MANNING @TKO     TAMOTSU MATSUI @OSA ETIENNE MICHELON @E 
ROGER MOORE @GEO    HERMAN OGGEL @GEO   JIM PEARSE @TRO     RACHEL PONG @HGO    
ROGER RAMASWAMY @AK DOBRI RASCHEEFF @TH PATRICK ROACH @VBO  AMICHAI SOBOL @ISO  
KANNANKOTE SRIKANTH OVE STAHL @GEC      CARLO TARANTOLA @VB ROGER THOMPSON @TKO 
TOMOYO TOYOSHIMA @O TOSHIO UEDA @TKO    IGOR VARSEK @KAR    HENK VERMAAT @ERO   
JUERGEN WEGWART@GMO THYGE WELLER @RTO