Natural language (NL) interfaces for database query formulation have always been recognized as a much needed enhancement for end-users. Poor performances with earlier NL systems had led to a lull in research in this field. However, latter-day experiments and systems appear to be sufficiently more promising to warrant continued and further research in this area. This paper proposes an experimental query interface that filters NL query statements for search predicates that are derived from constructs on conceptual schemas, thereby avoiding the computational difficulty with full-fledged NL parsing. A prototype of the Conceptual Query Language (CQL) exists as a front-end to an Oracle relational DBMS.

Natural language query filtration in the Conceptual Query Language

The paper presents an automatic acquisition of linguistic patterns that can be used for knowledge based information extraction from texts. In knowledge based information extraction, linguistic patterns play a central role in the recognition and classification of input texts. Although the knowledge based approach has been proved effective for information extraction on limited domains, there are difficulties in construction of a large number of domain specific linguistic patterns. Manual creation of patterns is time consuming and error prone, even for a small application domain. To solve the scalability and the portability problem, an automatic acquisition of patterns must be provided. We present the PALKA (Parallel Automatic Linguistic Knowledge Acquisition) system that acquires linguistic patterns from a set of domain specific training texts and their desired outputs. A specialized representation of patterns called FP structures has been defined. Patterns are constructed in the form of FP structures from training texts, and the acquired patterns are tuned further through the generalization of semantic constraints. Inductive learning mechanism is applied in the generalization step. The PALKA system has been used to generate patterns for our information extraction system developed for the fourth Message Understanding Conference (MUC-4). >

Acquisition of linguistic patterns for knowledge-based information extraction

Scheme for the automated apprehension of textual information conveyed in an input string. The input string is segmented to generate segments and/or semantical units. The following steps are repeated for each segment in the input string until a subset for each segment in said input string is identified:
 a. identifying a matching semantical unit in a fractal hierarchical knowledge database of semantical units and pointers, said matching semantical units being deemed to be related to a segment of said input string, b. determining the fitness of said matching semantical unit by taking into consideration said semantical unit's associations, c. defining a subset of information related to said matching semantical unit within said fractal hierarchical knowledge database. Then these subsets are combined to form a resulting semantic network.

Processing of textual information and automated apprehension of information

A knowledge acquisition tool to extract semantic patterns for a memory-based information retrieval system is presented. The major goal of this tool is to facilitate the construction of a large knowledge base of semantic patterns. The system acquires semantic patterns from texts with a small amount of user interaction. It acquires new phrasal patterns from the input text, maps each element of the pattern to a meaning frame, generalizes the acquired pattern, and merges it into the current knowledge base. Interaction with the user is introduced at some decision points, where the ambiguity cannot be resolved automatically without other pieces of predefined knowledge. The acquisition process is described in detail, and a preliminary experimental result is discussed. >

Acquisition of semantic patterns for information extraction from corpora

This dissertation presents several new methods of supervised and unsupervised learning of word sense disambiguation models. The supervised methods focus on performing model searches through a space of probabilistic models, and the unsupervised methods rely on the use of Gibbs Sampling and the Expectation Maximization (EM) algorithm. In both the supervised and unsupervised case, the Naive Bayesian model is found to perform well. An explanation for this success is presented in terms of learning rates and bias-variance decompositions.

Learning Probabilistic Models of Word Sense Disambiguation

It is argued that a viable solution for building future intelligent systems is to design special-purpose parallel computer architectures. The applications are restricted to those using semantic networks for knowledge representation. Reasoning on these networks is achieved with a marker-passing model of processing. The Semantic Network Array Processor (SNAP), a marker-passing parallel computer dedicated for natural-language and other knowledge-processing applications, is considered. Solutions for several nontrivial natural-language problems using the marker-passing approach are discussed. >

M. Chung

Papers

SNAP: parallel processing applied to AI