Semantic Textual Similarity

709 Words3 Pages
Abstract— Semantic textual similarity(STS) assess the degree of semantic similarity between pair of sentences. Every sentence is built using syntactic rules and semantic relations. To estimate the similarity there is a need to generate syntactic and semantic features. These features are combined using regression models. Semantic features will only deal with the meaning of individual words between the sentence pair. So, by considering only semantic features similarity among sentences cannot be estimated. For constructing a sentence out of these individual words syntactic features are required. In this paper, proposing phrase entity a new syntactic feature. The experimental work is carried out showing the importance of syntactic features…show more content…
It plays a vital role in day to day life. Mainly the information/communication is carried out using text. Sharing information through the internet has become so popular and it became daily routine. So, Information in the form of text is growing rapidly. To search for the most relevant information there is a need to find the similarity between the text. But in NL, the text can be expressed in various ways without changing the meaning of a sentence. Therefore, finding the similarity between the text snippets became a research problem. The similarity can be measured among words, phrases, sentences, paragraphs, documents and word to phrase, word to sentence, sentence to paragraph etc.

Techniques used for measuring the similarity between the long texts (documents) and short texts (sentences) are different. Finding the similarity between the long texts mainly depends on the number of common words they are sharing whereas common words between the short texts can be rare or null. For example, the bird bathing in sink and Birdie is washing itself in the water basin are the two sentences which are equivalent but have less number of common words.
So, measuring the similarity between the sentences is critical. In all the languages, every sentence is built following syntax rules and semantic relations. For measuring the similarity between sentences, syntactic and semantic features need to be
…show more content…
Applications such as 1) Text summarization: STS is used in grouping of semantic similar sentences [1], 2) Machine translation evaluation: STS is used to measure the degree of equivalence between the machine generated translation and the referenced translation [2], 3) Information retrieval: STS is used for measuring the semantic equivalence between pairs of texts [3], 4) Web page retrieval: STS is used by measuring the page title similarity [4] and also in many systems such as plagiarism detection, question-answer

More about Semantic Textual Similarity

Open Document