The “Jump and Stay” Method to Discover Proper Verb Centered Constructions in Corpus Lattices

19693

Authors

nan

Abstract

AbstractThe research presented here is based on the theoretical model of corpus lattices. We implemented this as an effective data structure, and developed an algorithm based on this structure to discover essential verbal expressions from corpus data. The idea behind the algorithm is the “jump and stay” principle, which tells us that our target expressions will be found at such places in the lattice where the value of a suitable function (defined on the vertex set of the corpus lattice) significantly increases (jumps) and then remains the same (stays). We evaluated our method on Hungarian data. Evaluation shows that about 75% of the obtained expressions are correct, actual errors are rare. Thus, this paper is 1. a proof of concept concerning the corpus lattice model, opening the way to investigate this structure further through our implementation; and 2. a proof of concept of the “jump and stay” idea and the algorithm itself, opening the way to apply it further, e.g. for other languages.

Keywords

nan

Full Content

Proceedings of Recent Advances in Natural Language Processing, pages 1076–1084, Varna, Bulgaria, Sep 2–4, 2019. https://doi.org/10.26615/978-954-452-056-4_124 1076 The “Jump and Stay” Method to Discover Proper Verb Centered Constructions in Corpus Lattices Bálint Sass Research Institute for Linguistics, Hungarian Academy of Sciences Budapest, Hungary sass.balint@nytud.mta.hu Abstract The research presented here is based on the theoretical model of corpus lattices. We implemented this as an effective data structure, and developed an algorithm based on this structure to discover essen- tial verbal expressions from corpus data. The idea behind the algorithm is the “jump and stay” principle, which tells us that our target expressions will be found at such places in the lattice where the value of a suitably defined function (whose domain is the vertex set of the corpus lattice) sig- nificantly increases (jumps) and then re- mains the same (stays). We evaluated our method on Hungarian data. Evaluation shows that about 75% of the obtained ex- pressions are correct, actual errors are rare. Thus, this paper is 1. a proof of concept concerning the corpus lattice model, open- ing the way to investigate this structure further through our implementation; and 2. a proof of concept of the “jump and stay” idea and the algorithm itself, open- ing the way to apply it further, e.g. for other languages. 1 Introduction In this paper we present a novel, original verbal construction discovery method. Our starting point will be our former paper (Sass, 2018) which de- scribes a theoretical model considered an appro- priate basis for extracting so-called proper verb centered constructions from analysed corpora. First, let us look at our target. In this terminol- ogy, verb centered constructions (VCC) are verb + slot structures, where slots can be unfilled (free) or filled (by a filler word). In English, subject (SBJ), direct object (OBJ) and all prepositions can be considered as slots. For example, ‘take + SBJ + OBJ + into:account’ has two free slots (SBJ and OBJ) and a filled ‘into’ slot where the filler (marked by a colon) is the word ‘account’. In this approach, a filler is the head of the phrase realizes the slot. Length of a VCC (l) is defined as number of slots and fillers added up, the above example has a length of 4. Then, what are proper verb centered construc- tions (pVCC)? They are complete and clean. That means they contain all necessary elements, and does not contain any unnecessary element for ex- pressing the core meaning of the verbal expression in question. For example, ‘take + SBJ + OBJ:part + in’ is proper, while ‘take + SBJ + OBJ:part’ is not proper (because not complete), and ‘read + SBJ + OBJ’ is proper, while ‘read + SBJ + OBJ:book’ is not proper (because not clean). In other words, free slots in pVCCs are complements (subject included) and fillers in pVCCs are id- iomatic, carrying some special meaning. It is clear that pVCCs are constructions. They are form–meaning pairs (Goldberg, 2006; Kay and Michaelis, 2015), they …

Content truncated to 500 words

Labeled Text

AbstractThe research presented here is based on the theoretical model of corpus lattices. We implemented this as an effective data structure, and developed an algorithm based on this structure to discover essential verbal expressions from corpus data. The idea behind the algorithm is the “jump and stay” principle, which tells us that our target expressions will be found at such places in the lattice where the value of a suitable function (defined on the vertex set of the corpus lattice) significantly increases (jumps) and then remains the same (stays). We evaluated our method on Hungarian data. Evaluation shows that about 75% of the obtained expressions are correct, actual errors are rare. Thus, this paper is 1. a proof of concept concerning the corpus lattice model, opening the way to investigate this structure further through our implementation; and 2. a proof of concept of the “jump and stay” idea and the algorithm itself, opening the way to apply it further, e.g. for other languages.

Publication Details

Paper ID:
19693
Year:
2019
Publication:
TIPSTER
Publication Type:
conf
DOI:
nan