Home | Biodata | Biography | Photo Gallery | Publications | Tributes
[Back to Epigraphy List]

Epigraphy


Developing an expert system for Indian epigraphy
 Gift Siromoney, Chandrasekaran R. and Suresh D., 1985

Abstract

An expert system is a computer program which can be consulted by a layman to get some kind of expert advice in a limited area of knowledge. With the availability of computers with large memory it has been possible to demonstrate the usefulness of expert systems in medical diagnosis and other fields. An expert system in archaeology has been reported from France.

We suggest that we should develop expert systems which will be of use to scholars who are working in the area of epigraphy. The state of art is such that it is possible to build such an expert system. The system can be built in different modules.

A module can be developed for each Indian script and the module can be used in dating of inscriptions on the basis of paleography. A second module can be built for finding dates using astronomical data. A third module can be constructed based on the contents of inscriptions and work can be started using personal names compiled by Karashima's concordance.

We have developed computer programs that can be used to date medieval Tamil inscriptions on the basis of paleography and it is proposed to demonstrate the program at the Epigraphical Congress.

Introduction

The concept of an expert system has emerged from the relatively new field of artificial intelligence1 or AI. The term expert system designates a system that is an expert in some existing human art and the term also designates a system that plays the role of a consultant. An expert system is a computer program which can be made use of even by a layman to get some kind of expert advice in a limited area of knowledge.

To build an expert system in a specific field, experts in that field are consulted and the essence of the expert knowledge is extracted by talking to the experts. This knowledge is converted into a form that is suitable for use by the computer, and this kind of work is called knowledge engineering.

Expert systems have been developed in areas such as medical diagnosis2 and recently an expert system in archaeology3 has been reported from France.

One of the aims of this paper is to stimulate the interest of Indian epigraphists so that we could develop expert systems that will be of use to scholars who are working in the area of Indian epigraphy. The current state of the art in the world is such that expert systems can be built in India.

Examples of Expert Systems:

Example 1

One of the well known expert systems is called MYCIN2 and it was developed at Stanford University. MYCIN is viewed as a clear example of the experimental aspect of artificial intelligence. It is an expert system for diagnosing infectious diseases. The term MYCIN ( like chloromycin, entromycin, etc) is a common suffix associated with many antimicrobial agents and it was given as the name of that project. At the initial stages MYCIN was planned as

a) consultation program, a program to provide doctors with advice regarding antimicrobial therapy using both clinical data and judgemental decisions,
b) one with interactive explanation capabilities which will explain and justify its recommendations and
c) one which can acquire new judgemental knowledge from experts in the field.

MYCIN learns about particular cases by requesting information about the patient. The question and answer session will go something like this.

Computer: What is the patients' name?
       User: Bruce Winston.
Computer: Male or Female?
       User: Male.
Computer: Age?
       User: He is 45.
. . .. . .. . . . . .
. .. . . .. . . . .. . .
Computer: how much does the patient weigh?
       User: 70 kilograms.
Computer: My recommendation is as follows: Give epigramycin using a dose of 119 mg (1.7 mg/kg for   10 days.  If diarrhea develops check for…. ….

Example 2

A recent work on vegetable gardening expert sytem4 called VEGE has the following typical question answer session.
        User: When should tomato plants be started indoors?
Computer: Your tomatoes should be planted indoors from April to May.
       User: Will pumpkin plants sprout after being planted 4 inches deep?
Computer: No. Pumpkins should be planted about 1 inch deep.
      User:  How many eggplants will fit into a 50 feet row?
Computer: You can fit from 25 to 35 eggplant plants in a row that size.

We have given these as illustrations to show how expert systems are also in many cases query answering systems.

Expert systems for Indian Epigraphy:

Expert systems are often developed in different modules. We need to develop separate modules from the different regions and scripts and the modules can be integrated into a single system. A separate module can be developed for finding dates using astronomical data found in inscriptions. Another module can be constructed on the basis of the contents of inscriptions. For Chola inscriptions in Tamil, one can start with the personal names compiled by Karashima et al.5

Interactive system for dating Tamil Inscriptions:

We have developed a computer program at the Madras Christian College, that can be used to assign probable dates to medieval Tamil inscriptions. The work is at an experimental stage but we propose to demonstrate at the Twelfth Congress of the Epigraphical Society of India, how it works. Our system is based on paleography and we have yet to incorporate the query answering procedure using simple English.

The method used can be summarized as follows. First, authentic inscriptions are collected whose dates are already known with certainty. By date we mean the date of the time of inscribing (i.e. copying) of the inscriptions and not necessarily the time of any donation mentioned in the inscription. The larger the number of inscriptions the better. Secondly paleographic features are extracted from different letters from each authentic inscription. Thirdly features from each inscription are encoded in numerical form. Thus each inscription is represented as a number, depending upon whether a feature is present or absent from an inscription. The authentic inscriptions are stored in the computer in the form of a library. Fourthly a given unknown inscription or test inscription, is encoded. Finally the encoded form of the test inscription is matched with the library inscriptions, using some method and the test inscription is given a date on the basis of the comparison made with the library inscriptions.6,7,8

We present here the information given at different stages by the computer in our program. The user has to supply the information to the computer by typing in the necessary information. Here are the computer messages:


  1. COMPUTER ASSISTED DATING OF
    MEDIEVAL TAMIL INSCRIPTIONS DEVELOPED AT 
    MADRAS CHRISTIAN COLLEGE, TAMBARAM, MADRAS 600 059.
  2. Keep before you a copy of the inscription you wish to date, using this program. We shall call this inscription "TEST INSCRIPTION". When you are ready for the next step, please press the -(Return)-key. 
  3. The computer will display character after character. Each character will be displayed in different shapes or styles. Please select the style closest to what you have in the "TEST INSCRIPTION". You have to do this for a total of 32 important characters.
    When you are ready for the next step please press the -(RETURN)-key.
  4. Here is a simple procedure.
    1. Observe carefully the different shapes of a character that will be displayed.
    2. Please identify the shape closest to the one occurring in your "TEST INSCRIPTION".
    3. Note that each shape in the display has a number.
    4. Please note the number of the selected character.
    5. You will have to type in that number, when requested to do so, after the display is over.
    6. If you do not find a match between what is displayed and what you have in the "TEST INSCRIPTION" please type the number ' 0 '.
      When you are ready for the next step please press the -(RETURN)-key.
  5. The screen can accommodate a maximum of twelve shapes. The screen either displays all the various shapes of a character or the maximum it can accommodate. If you finished viewing the screen press the -(RETURN)-key.
    When you are ready for the next step, please press the -(RETURN)-key.
  6. If you want to read the procedure once again please type PROCEDURE REQUESTED otherwise press the -(RETURN)-key.
  7. Please wait, Computer is now reading the information and storing it in its memory. Please wait.
  8. The test inscription is assigned to the period:
    --- --- to --- --- A.D.
  9. The test inscription belongs to one of the following periods:
    ------ to ---- --- A.D.
    ------ to ---- --- A.D.
  10. Warning: spurious data, incomplete data or erroneously recognized features may lead to ties.
Conclusion

We have demonstrated that it is possible to develop and use computer programs that will be of assistance to the expert and novice alike for dating medieval Tamil inscriptions. More work is to be done to make the system robust and sensitive to take in even scribal handwriting features. We hope to incorporate query answering capability to the system. This is the time to develop expert systems for Indian epigraphy and the climate is just ripe in our country.

References:

1. Winston, P.H, Artificial Intelligence, Addison-Wesley, Reading, Second edition, 1979.
2. Buchanan, B.G,. and Shortliffe, H. Rule-based Expert Systems, Addison-Wesley, Reading, Second edition, 1984.
3. Lagrange, M., and Renaud, H, Intelligent knowledge-based systems in archaeology: a computerized simulation of reasoning by means of an expert system, "Computers and the Humanities", Vol.19, pp. 37-49, 1985
4. Lehnert, W.G., and Bain W.M.,VEGE : Variable processing in a natural language query system, Research Report 185, Department of Computer Science, Yale University, New Haven, Connecticut.
5. Karashima, N. Subbarayulu,Y., and Matsui,T., A Concordance of the Names of the Chola Inscriptions, Sarvodaya  Ilakkiya Pannai, Madurai, 1981.
6. Siromoney, G., Chandrasekaran, M., and Chandrasekaran, R., Computer methods of dating Tamil inscriptions, Proceedings of the Fifth International Conference-Seminar of Tamil Studies, Arunachalam, M., (Ed.) Madurai, 1981..
7. Chandrasekaran,R., Computer Recognition of Certain Indian Scripts, Doctoral Dissertation, University of Madras,1982.
8. Chandrasekaran, R., Chandrasekaran,M., and Siromoney, G., Computer dating of medieval inscriptions: South Indian Tamil, Computers and Humanities, Vol.17, pp.199-208, 1985.

Acknowledgement

We are grateful to Dr M. Chandrasekaran and  Messrs. R. Srinivasan and Abdul Huq who contributed extensively to the collection and organization of data at the initial stages of the work. The program was developed in BASIC at the Kibble Center for Statistical Computing at the Department of Statistics, Madras Christian College on a Genie-1 Microcomputer.

Go to the top of the page

Home | Biodata | Biography | Photo Gallery | Publications | Tributes