Applying a lightweight iterative merging chinese segmentation in web image annotation
MLDM'13 Proceedings of the 9th international conference on Machine Learning and Data Mining in Pattern Recognition
| Hi-index | 0.00 |
In this paper we propose a new method of linear textsegmentation based on lexical cohesion of a text. The usualsteps ( to compute the lexical chains according to relatednesscriteria, to score the chains after different parameters,to select the strong chains, to obtain the segments) arereplaced by a single procedure. Namely, a single chain ofdisambiguated words in a text is established and the ripsof this single chain are considered as boundaries of thesegments of the cohesion structure of the text (CohesionTextTiling or CTT).The summaries of arbitrarily length are obtained by extractionusing three different methods applied to the obtainedsegments. The informativeness of the obtained summaries iscompared with the informativeness of the pair summaries ofthe same length obtained using an earlier method of logicalsegmentation (coherence segmentation) by text entailment(Logical TextTiling or LTT). Some experiments about CTTand LTT methods are made for four ”classical” texts insummarization literature. The conclusion is that the qualityof the summarization using cohesion segmentation (CTT) isbetter than the quality using logical (coherence) segmentation(LTT).