Case Base Reasoning
XMIS (XML query language for Mediator and Integrator Systems)
Since CBR stores actual experiences, one immediate consequence is its deployment as a tool to build a corporate memory. Every experience encountered is stored as a case, and the case base becomes a repository of all experiences. On encountering a problem needing a solution, the CBR system uses inexact matching techniques to retrieve from memory a set of similar experiences, in the process also producing possible solutions to the present problem. In this way CBR can enhance a database to serve as a knowledge base.
One of the earliest attempts by the AI&DB lab was to build a tool to support the launch campaign at ISRO, SHAR. The campaign to assemble a launch vehicle typically lasts a few months and involves many people working together. The first part of the project constituted a user interface to monitor the different activities, following the Gantt chart. It also used a genetic algorithm based scheduler to produce a revised scheduled when needed. The second part was a tool to build a repository of problems encountered and solutions found. The aim was that such a repository would grow with experience across all launch vehicle assembly campaigns. A prototype system was built.
An ongoing project for Carborundum Universals (CUMI) highlights another aspect of CBR. That is its applicability in situations where analytical knowledge is incomplete and yet experienced personnel can make decisions that succeed. The accompanying flowchart (courtesy CUMI) depicts the refractory making process employed by the company. The process starts with design; followed by pattern making, mould making and flasking of the mould. Meanwhile the appropriate raw material is mixed. The mix is melted in an electrocast furnace and poured into the mould. Deflasking, finishing and inspection follow a long period of annealing. The production processes has many parameters and takes many days to be completed. It is not always easy to pinpoint the cause when a manufactured part is defective. Accumulated experience however enables some experts to decide upon the process modification needed. Our project with them involves setting up a CBR system on the floor, collecting operational data, and maintaining a repository of all kinds of blocks. Each block manufactured, good or bad, is added to the memory along with the corrective action taken if any. When a defective block is encountered the CBR system retrieves the best matching corrective action. It is hoped that with sufficient data accumulating, the system will, in the future, be in a position to support predictive maintenance and trigger changes in the design handbook.
Another application where archived data could be profitably used is in the recently sanctioned ground based autonomy project for ISRO, Master Control Facility. This project is part of an endeavor to make the satellite control activity less human intensive. A large part of human attention is directed towards monitoring a few hundred telemetry parameters describing the current health of the satellite. Our project is to mine the past data to build systems to support health assessment. The objective is to recognize pattern combinations that are indicative of an abnormality, and send an alert to a human expert.
AI Planning is a sub-field of AI which focuses on the development of algorithms that can model such goal-oriented behavior by synthesizing a course of action that can achieve the desired objectives. These algorithms serve the needs of intelligent agents that require some kind of independent, autonomous existence capability.
An exciting project with ISRO, Satellite Centre, is working towards onboard autonomy of a low earth orbiting satellite. Due to the fact that these satellites fly in orbits closer to earth, their visibility is only for relatively short durations of time. In these periods of visibility currently ground control needs to handle both payload scheduling and health monitoring. The objective is again to work towards decreased human dependence (see accompanying figure). The work involves planning and scheduling, with the onboard system itself making decisions on initiating the necessary processes. A major challenge is the need for robustness and reliability under severe computing resource limitations. It is envisaged that this work will involve the application of emerging first principles planning techniques.
Apart from the work described above the AI&DB lab has done work in the following areas: Natural language generation, with the aim of building articulate systems that will be able to give instructions or explanations in human languages. This could be for pedagogical systems, and also for retrieving and conveying information from databases, for example medical histories, and processes for example weather forecasting: Data warehousing and mining, to extract useful information from large amounts of typically operational data: And we are actively looking at XML based semi-structured databases that will be useful in building heterogeneous knowledge bases with wider access.
The Web has undoubtedly become the prime medium for information delivery, information collection, electronic commerce applications, distance education and a host of other applications. Web data is a mix of structured, semi-structured and unstructured data. eXtensible Mark-up Language (XML) is fast replacing the traditional Hypertext Mark-up Language (HTML) as the data exchange standard. The unique feature of semi-structured data is that the data is self-describing. The data contains the schema or structure information as part of the data. Several interesting issues such as storage of semi-structured data, devising query languages and efficient query processing in such databases that arise in this context are under active investigation in the AI-DB laboratory.
Research Projects in XML
A declarative specification based method for translating data between different XML representations has been devised. This method also helps in translating of queries in an information integration system. A general-purpose semi-structured data repository, called FLEXIS, capable of gathering, storing and efficiently retrieving XML data has been developed. The activities on XML data have been supported by a grant from MHRD.
A new XML query language, called XMIS (XML query language for Mediator and Integrator Systems), designed specifically for use in data integration systems has been developed. XMIS is rule based query language inspired by Datalog. It does not allow the creation of new element tags in the result document and thus preserves the original document structure. We have developed techniques for query rewriting for XMIS queries. A “publish-subscribe system” for XML data has been developed where XMIS is used as the subscription language. This system exploits query rewriting to efficiently serve a piece of XML data (or parts there of) to a group of subscribers interested in that data.
Query rewriting techniques have applications in Semantic Caching systems. For XML queries expressed in XMIS, we have designed and developed a semantic query cache and have experimentally shown that significant improvement in time taken for answering a query can be realized. In a Semantic Cache, queries along with their results are stored in a local store so that future queries can be efficiently answered. For a new query, we check if the whole query or any part of it can be answered from the cached queries. For a significant subset of the XML query language XML-QL also, we have been able to devise a method of query rewriting and then a semantic cache system.
We have also developed a Document Type Descriptor (DTD) guided indexing mechanism for indexing XML documents so that an efficient Web search engine for XML documents can be constructed.
The AI&DB lab has also interacted with other groups, With the Model Based and Qualitative Reasoning group of TU-Munich, Dept of Chemical Engineering- IITM, and MS Swaminathan Research Foundation- Chennai. We have explored building qualitative models for diagnosis to support environmental decision making. A system for Model Based Resource Management Aid (Mermaid) was designed, and a prototype system to model a part of the mangrove forest in Pichavaram, Tamilnadu, was built. In the area of CBR we are interacting with the AI group TRDDC, Pune, with plans for increased collaboration, to work towards building knowledge management tools for both personal and organizational use.