A Frame-based Approach to Text Generation

A Frame-based Approach to Text Generation* Huong Thanh Le Faculty of Information Technology, Hanoi University of Technology 1 Dai Co Viet street, Hanoi, Vietnam huonglt@it-hut.edu.vn Abstract. This paper is a study on constructing a natural language interface to database, concentrating on generating textual answers. TGEN, a system that generates textual answer from query result tables is presented. The TGEN architecture guarantees its portability across domains. A combination of a frame-based approach and natural language generation techniques in the TGEN provides text fluency and text flexibility. The implementation result shows that this approach is feasible while a deep NLG approach is still far to be reached. Keywords: natural language interface, text generation, frame-based approach. 1. Introduction Database management systems (DBMSs) have been widely used thanks to their efficiency in storing and retrieving data. However, retrieving information from database requires users to compose queries in a query language (e.g., QBE or SQL) or to fill some search criteria in the interface (Liu, 1995; Catarci et al., 1997). In addition, traditional DBMSs are still limited by its capability to generate outputs. They normally dump query results in a table or a pre-defined form without any understanding of the meaning of data. The research on natural language interface to databases has recently received attention from the research communities (Wang et al., 1999; ELF Software Co., 2001; Torgersson and Falkman, 2002; Hallett et al., 2005; Bertomeu, 2006). The purpose of such a natural language interface is to allow users to compose queries in natural language and to receive responses under the form of short answers. The natural language interface is thus preferred than the traditional interface. A typical natural language interface has to solve two main tasks: (i) translating a natural language query to a query language; and (ii) generating a natural language answer by using information from a query result table. In this paper, we focus on solving the second task. A frame-based approach, which bases on predefined elementary frames and text generation rules to generate new frames and to produce flexible text, is introduced. The implemented system is called Text Generator or TGEN for short. The remaining sections of this paper are organized as follows: Section 2 introduces the architecture of our proposed Natural Language Interface for Querying Database and Automatically Generating Reports (NLI4DB), in order to get an overview of the TGEN role in the interface. Section 3 describes the rule set used by the TGEN to generate text. Section 4 * The author gratefully acknowledges the receipt of a grant from the Flemish Interuniversity Council for University Development Cooperation (VLIR UOS) which enabled the research team to carry out this work. Copyright 2007 by Huong Thanh Le 192 presents the major components and the data flow in the TGEN. In Section 5, two examples are given to illustrate the working process of the TGEN. Our implementation discussion and some experimental results are given in Section 6. Finally, Section 7 concludes the paper and proposes possible future work on this approach. 2. …

Content truncated to 500 words