Dissertation Information for Ladda Yindeemak
- Ladda Yindeemak
- Library and Information Science
- Indiana University (USA) (1988)
- Stephen P. Harter
- David Kaser
- Vichit Lorchirachoonkul
- John F. Hartmann
MPACT Status: Fully Complete
Title: Computer processing with Thai text: Keyword in context indexing
Abstract: An information system which is involved with text analysis generally deals with indexing and searching processes. Overall, automatic indexing employs either a statistical or linguistic approach in assigning appropriate terms capable of representing the content of the document. The need to apply computer technology to process information in Thai materials is imperative. However, processing Thai text with the computer faces many difficulties due to the characteristics of the Thai language. The main problems are: (1) words in Thai are mostly compound words whose word boundaries cannot be clearly defined, and the compound words may contain common function words; (2) written Thai does not use interword spaces and seldom uses punctuation marks, in contrast to the writing of most Western languages.
The objective of this research is to develop and evaluate an automated system for producing Thai keyword in context indexing based on word-frequency methods. Due to the problem of identifying word boundaries, all Thai words are identified as substrings of words which are called "tokens". An algorithm for recognizing tokens which was developed at the National Institute of Development Administration (NIDA) in Thailand was employed. A stopword method was used to eliminate the function words that were considered not to be indicative of the content matter. The frequency techniques used for indexing were: (1) the frequency of token occurrence; and (2) the "ad hoc" neighborhood frequency of token occurrence, used to assign the concatenations of tokens that appear as neighbors into words or phrases. Absolute frequency with further modification was used to weight the significant words for selecting keywords describing a given document.
The evaluation of automatically-assigned keywords was performed by comparing them to manually-assigned keywords. Recall and precision were the criteria used to measure the effectiveness of the indexing system. The results of the automatic indexing system were found to be satisfactory and encouraging. The indexing techniques, the factors affecting the system performance, and suggestions for further study are discussed. Many studies can be built upon this research.
MPACT Scores for Ladda Yindeemak
A = 0
C = 0
A+C = 0
T = 0
G = 0
W = 0
TD = 0
TA = 0
calculated 2008-04-20 11:56:12
Advisors and Advisees Graph