Neural Networks for Natural Language Processing by S. Sumathi

Author:S. Sumathi
Language: eng
Format: epub
Publisher: Engineering Science Reference

The semantic similarity between the concepts considers mainly on the syntactic structure. The similarities between the sentences are considered by calculating the sum of the similarities between the head of the sentence. It is also important to note that the gloss-based measures and HSO are the only measures that can be computed to find the similarity between the two words that are used in different Parts Of Speech(POS).

(3.1)

Let the sentence S1 is made of n phrases and their heads are h11,…, h1n and sentence S2 is made of n phrases, and h21,…, h2n are their heads. The phrases of h1i and h2i probably have the same syntactic function. The syntactic role for one of the sentence is considered. A Penalisation Factor (PF) is introduced in case if one sentence phrase is not shared by the other sentence to reflect the fact that one of the sentences has extra information. If the heads are not present in WordNet like pronouns, they are ignored in the calculation unless the same word shares the same syntactic role in both the sentences. HSO method overcomes even in the absence of proper nouns in WordNet. In the proposed RLQC method, the penalisation factor is eliminated for calculating the semantic similarity between the questions. HSO uses the syntactic role of both the sentences, so that PF is not needed for calculating the similarity.

(3.2)

Let the question q1 is made of n phrases and their heads are h11,…, h1n and question q2 is made of n phrases, and h21,…, h2n are their heads; and moreover the phrases of h1i and h2i have the same syntactic function.

EXPERIMENTS AND RESULTS

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.