| My research is focused on two important issues: whether thesauri enhance retrieval effectiveness and whether automatic indexing can compete with manual indexing in a Chinese information retrieval system.; An interactive Chinese information retrieval system named CIRS was built for these experiments. 555 abstracts in Chinese from ko-chi-chien-shiunn published by the Science and Technology Information Center, Republic of China and 30 queries were used in my experiments. A relational thesaurus, a supplementary resource for users, was built to be interactive. Two indexing methods, automatic indexing and manual indexing, are supported in the system. The User Interface in the system provides users with the functions to construct queries, execute queries, and view the titles and the abstracts of the retrieved documents. A query is an array of 56 cells where keywords or operators can be entered. AND, OR, NOT, Left and Right Parentheses are five operator choices for the query construction. To construct a query, users can enter one or more leading words so that a list of keywords matching such leading words appear for selection. The selected keyword can lead to the display of related keywords selected from the relational thesaurus if users desire to further clarify the intended meaning of their query. In addition, users are also allowed to view and reuse the previously selected keywords.; Recall, precision, and two nonparametric statistical tests are used to measure and evaluate the effectiveness of the system. We examined three hypotheses: that the retrieval effectiveness with the thesaurus is better than that without the thesaurus in the automatic indexing or in the manual indexing environment and that the retrieval effectiveness of the system with automatic indexing is as least as good as that given by the system with manual indexing.; Statistical analysis of the recall and precision measure indicate that the relational thesaurus does improve the retrieval effectiveness both in the automatic indexing environment and in the manual indexing environment and that automatic indexing is at least as good as manual indexing. |
