Martha W. Evens

SEMANTICALLY SIGNIFICANT PATTERNS IN DICTIONARY DEFINITIONS

Natural language processing systems need large lexicons containing explicit information about lexical-semantic relationships, selection restrictions, and verb categories. Because the labor involved in constructing such lexicons by hand is overwhelming, we have been trying to construct lexical entries automatically from information available in the machine-readable version of Websters Seventh Collegiate Dictionary. This work is rich in implicit information; the problem is to make it explicit. This paper describes methods for finding taxonomy and set-membership relationships, recognizing nouns that ordinarily represent human beings, and identifying active and stative verbs and adjectives.

Comparing words, stems, and roots as index terms in an Arabic Information Retrieval System

The Micro‐AIRS System, a microcomputer system for Arabic Information Retrieval, was designed as an experimental system to investigate indexing and retrieval processes for Arabic bibliographic data. A series of experiments were performed using 29 queries against a base of 355 Arabic bibliographic records, covering computer and information science from the bibliographic databank at King Abdulaziz City for Science and Technology. These experiments revealed that using roots and using stems as index terms gives better retrieval results than using words. The root performs as well as or better than the stem at low recall levels and definitely better at high recall levels. Several different binary similarity coefficients were tried: the cosine, Dice, and Jaccard coefficients. All three led to exactly the same document rankings for every query. The experiments were run on an IBM/AT‐compatible microcomputer. Micro‐AIRS is written in Turbo C, Version 2.0.

Relational thesauri in information retrieval

This article describes the design and development of a new type of thesaurus based on lexical‐semantic relations. Relational thesauri have been constructed to perform a new type of term classification. These relational thesauri are generally applicable to any document collection and their maintenance is relatively simple. A series of experiments to evaluate thesauri of this new type have been run on an information retrieval system called IRS at Illinois Institute of Technology. The results of experiments with queries enhanced using thesauri based on several different groups of relations have been compared against performance with the original queries. Thesauri from most groups, except antonyms, made improvements in recall as well as in precision. The best results come from a set of ill‐formed queries with few index terms. These results have been analyzed with both precision‐recall graphs and statistical tests. While thesauri that combine a number of relations together have been most effective in a batch environment, there is reason to believe that individual relations will be more useful in an interactive retrieval system that presents index terms to the user and allows him to choose those that best convey his meaning.

A computational morphology system for Arabic

This paper describes a new algorithm for morphological analysis of Arabic words, which has been tested on a corpus of 242 abstracts from the Saudi Arabian National Computer Conference . It runs an order of magnitude faster than other algorithms in the literature.

CIRCSIM-Tutor: An Intelligent Tutoring System Using Natural Language Dialogue

CIRCSIM-Tutor version 2, a dialogue-based intelligent tutoring system (ITS), is nearly five years old. It conducts a conversation with a student to help the student learn to solve a class of problems in cardiovascular physiology dealing with the regulation of blood pressure. It uses natural language for both input and output, and can handle a variety of syntactic constructions and lexical items, including sentence fragments and misspelled words.

Stemming methodologies over individual query words for an Arabic information retrieval system

Stemming is one of the most important factors that affect the performance of information retrieval systems. This article investigates how to improve the performance of an Arabic Information Retrieval System (Arabic-IRS) by imposing the retrieval method over individual words of a query depending on the importance of the WORD, the STEM, or the ROOT of the query terms in the database. This method, called Mixed Stemming, computes term importance using a weighting scheme that uses the Term Frequency (TF) and the Inverse Document-Frequency (IDF), called TFxIDF. An extended version of the Arabic-IRS system is designed, implemented, and evaluated to reduce the number of irrelevant documents retrieved. The results of the experiment suggest that the proposed method outperforms the Word index method using the Binary scheme and the Word index method using the TFxIDF weighting scheme. It also outperforms the Stem index method using the Binary weighting scheme but does not outperform the Stem index method using the TFxIDF weighting scheme, and again it outperforms the Root index method using the Binary weighting scheme but does not outperform the Root index method using the TFxIDF weighting scheme.

BUILDING A LARGE THESAURUS FOR INFORMATION RETRIEVAL

Information retrieval systems that support searching of large textual databases are typically accessed by trained search intermediaries who provide assistance to end users in bridging the gap between the languages of authors and inquirers. We are building a thesaurus in the form of a large semantic network to support interactive query expansion and search by end users. Our lexicon is being built by analyzing and merging data from several large English dictionaries; testing of its value for retrieval is with the SMART and CODER systems.

Classifying Student Initiatives and Tutor Responses in Human Keyboard-to-Keyboard Tutoring Sessions

This study analyzed twenty-eight 1-hr-long tutoring sessions that were carried out keyboard-to-keyboard with tutor and student in different rooms. The tutors were professors of physiology at Rush Medical College. The students were 1st-year medical students. We classified student initiatives and tutor responses in human tutoring sessions with the goal of making our intelligent tutoring system capable of handling mixed-initiative dialogue. Student initiatives were classified along 4 dimensions: communicative goal, surface form, focus of attention, and degree of certainty (i.e., does the student hedge or not?). Student goals included request for confirmation, request for information, challenge, refusal to answer, and conversational repair. Tutor responses were classified along 3 dimensions: communicative goal, surface form, and delivery mode. The tutor goals included causal explanation, acknowledgment, conversational repair, instruction in the rules of the game, teaching the problem-solving algorithm, and teaching the language of physiology. Our interrater reliability studies supported these categories in the domain of tutoring.

Spelling Correction using Context

This paper describes a spelling correction system that functions as part of an intelligent tutor that carries on a natural language dialogue with its users. The process that searches the lexicon is adaptive as is the system filter, to speed up the process. The basis of our approach is the interaction between the parser and the spelling corrector. Alternative correction targets are fed back to the parser, which does a series of syntactic and semantic checks, based on the dialogue context, the sentence context, and the phrase context.

Circsim-tutor: an intelligent tutoring system for circulatory physiology

The aim of this research is to develop an intelligent tutoring system (ITS) which teaches students the causal relationships between the components of the circulatory physiology system and the complex behavior of the negative feedback system that stabilizes blood pressure. This system will accept natural language input from students and generate limited natural language explanations. It contains rules that identify the students errors and build a “bug-based” student model. It uses tutoring rules to plan each response based on its model of the student and the dialog history so that it can tailor the dialog to fit the students learning needs. The tutoring rule interpreter manages the dialog and determines strategy and tactics to achieve its educational goals.