A Grammar-Based Semantic Similarity Algorithm for Natural Language Sentences

Algorithm 2

Semantic sentence similarity.

INPUT: 𝑆_𝐴,  𝑆_𝐵, 𝜂 /* sets of relations of sentences A, B */

OUTPUT: 𝑆𝐼𝑀_𝐴𝐵

(1)   𝐿𝑇_𝐴← LinkingTypes(𝑆_𝐴, 𝜂)

(2)   𝐿𝑇_𝐵← LinkingTypes(𝑆_𝐵, 𝜂)

(3)  FOR ALL   𝜏∈𝐿𝑇_𝐴.type ∩𝐿𝑇_𝐵.type  DO

(4)   𝑆𝐼𝑀_𝐴𝐵← 𝑆𝐼𝑀_𝐴𝐵 + GrammarMatrix(𝐿𝑇_𝐴·𝜏, 𝐿𝑇_𝐵·𝜏)

(5)  END FOR

(6)   𝑆𝐼𝑀_𝐴𝐵← log𝑆𝐼𝑀_𝐴𝐵/𝐿𝑇_𝐴∩𝐿𝑇_𝐵

(7)  RETURN   𝑆𝐼𝑀_𝐴𝐵