Surface Realisation from Knowledge-Bases

Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics, pages 424–434, Baltimore, Maryland, USA, June 23-25 2014. c⃝2014 Association for Computational Linguistics Surface Realisation from Knowledge-Bases Bikash Gyawali Universit´e de Lorraine, LORIA Villers-l`es-Nancy, F-54600, France bikash.gyawali@loria.fr Claire Gardent CNRS, LORIA, UMR 7503 Vandoeuvre-l`es-Nancy, F-54500, France claire.gardent@loria.fr Abstract We present a simple, data-driven approach to generation from knowledge bases (KB). A key feature of this approach is that grammar induction is driven by the ex- tended domain of locality principle of TAG (Tree Adjoining Grammar); and that it takes into account both syntactic and semantic information. The resulting ex- tracted TAG includes a unification based semantics and can be used by an existing surface realiser to generate sentences from KB data. Experimental evaluation on the KBGen data shows that our model outper- forms a data-driven generate-and-rank ap- proach based on an automatically induced probabilistic grammar; and is comparable with a handcrafted symbolic approach. 1 Introduction In this paper we present a grammar based ap- proach for generating from knowledge bases (KB) which is linguistically principled and conceptually simple. A key feature of this approach is that grammar induction is driven by the extended do- main of locality principle of TAG (Tree Adjoining Grammar) and takes into account both syntactic and semantic information. The resulting extracted TAGs include a unification based semantics and can be used by an existing surface realiser to gen- erate sentences from KB data. To evaluate our approach, we use the bench- mark provided by the KBGen challenge (Banik et al., 2012; Banik et al., 2013), a challenge designed to evaluate generation from knowledge bases; where the input is a KB subset; and where the expected output is a complex sentence convey- ing the meaning represented by the input. When compared with two other systems having taken part in the KBGen challenge, our system outper- forms a data-driven, generate-and-rank approach based on an automatically induced probabilis- tic grammar; and produces results comparable to those obtained by a symbolic, rule based approach. Most importantly, we obtain these results using a general purpose approach that we believe is sim- pler and more transparent than current state of the art surface realisation systems generating from KB or DB data. 2 Related Work Our work is related to work on concept to text gen- eration. Earlier work on concept to text generation mainly focuses on generation from logical forms using rule-based methods. (Wang, 1980) uses hand-written rules to generate sentences from an extended predicate logic formalism; (Shieber et al., 1990) introduces a head-driven algorithm for generating from logical forms; (Kay, 1996) de- fines a chart based algorithm which enhances effi- ciency by minimising the number of semantically incomplete phrases being built; and (Shemtov, 1996) presents an extension of the chart based gen- eration algorithm presented in (Kay, 1996) which supports the generation of multiple paraphrases from underspecified semantic input. In all these approaches, grammar and lexicon are developed manually and it is assumed that the lexicon as- sociates semantic …

Content truncated to 500 words