Laboratory for Computational Linguistics

Probabilistic Morphological Analyzer for Hebrew Undotted Texts
Morphological analysis of words in a text is the first stage of most natural language applications. Due to the rich morphology of the Hebrew language and the inadequacy of the undotted script which results in a great degree of morphological ambiguity, the problem has not yet found a satisfactory solution. We notice that the problem of morphological analysis of Hebrew texts is similar to the well-studied problem of part-of-speech tagging in English, and thus can apply some of the approaches used to solve that problem. The work results in a morphological analysis which is correct for about 96% of the words. This result approaches results reported for English probabilistic part-of-speech tagging. It does so by using a very small training corpus - 5000 words only, in contrast to million-word corpora used for English tagging.
Researchers: Erel Segal, Alon Itai
Additional information

Corpus Based Analysis of Hebrew
The project's aim is to study ways to arrive at better tools for morphological and syntactic analysis of Hebrew using corpus based techniques.
The project is part of a joint project conducted in collaboration with the corpus based NLP group of the Computer Science Institute of of the Hebrew University, Jerusalem, headed by Prof. Eli Shamir. The project is sponsored by a grant from the Israel Ministry of Science.
Principle investigators: Alon Itai and Yoad Winter
Additional information.

Extensions and Implementations of Natural Logic:
This project develops an inference system for natural language that is based on Natural Logic: a logic that works directly on syntactic representations with no intermediate translation to a more familiar logical formalism. We use the proposal of Van Benthem (1987) and Sanchez (1991) as a starting point for a system that deals with monotonicity-based reasoning. This system is extended for natural language coordination and non-monotonic expressions in Categorial Grammar. Most of the recent developments in the project appear in an ICOS-2 paper.
The next stages of the project will involve a construction of a prototype system computing inferences in natural language, extensions to more linguistic phenomena and a general platform for extending various grammar formalism into (fragments of) natural logic.
The project is a joint enterprise of the Technion CS department and the UCLA Linguistics department, and it is sponsored by a grant from the Binationational Science Foundation (BSF).
Researchers: Nissim Francez, Yaroslav Fyodorov, Yoad Winter (Technion)
Ed Keenan, Henk Harkema, Shannon Madsen (UCLA)

Semantics of Natural Language Temporal Questions and Interfaces to Temporal Database Systems:
Traditional database systems retain snapshot information, valid at a given moment in time. Much research has been devoted to develop temporal databases which allow users to store time-dependent information. In recent years, there is an effort to consolidate and standardize this research into a single consensus temporal model and temporal query language, called SQL/Temporal.
SQL/Temporal adds several temporal constructs to standard SQL, and is therefore expected to be difficult to use for non-experts. We are developing a natural language interface for temporal databases, based on a semantic treatment of temporal questions. Using the interface, the user may express questions in natural language and have them automatically translated and submitted to the database. The project is sponsored by FIRST, administered by the Isreali Acedmy of science.
Researchers: Rani Nelken, Yoad Winter, Nissim Francez

Distributional Clustering using Large Corpora:
A system that constructs a knowledge base describing semantic similarity between texts by distributional clustering of expressions extracted from large coprpora using NLP techniques.
Researchers: Ron Bekkerman, Ran El-Yaniv, Yoad Winter

If you have comments or suggestions, email us at lcl@cs.technion.ac.il