Study On Sentence Similarity Algorithm For CBR-BDI Reasoning Mechanism

Under the influence of the industrial 4.0 environment,the robot industry is booming.In order to make the robot reach a higher level of intelligence,it needs an intelligent reasoning machine to reason,solve and study the real-time problem.Case-based reasoning is widely used in intelligent robots because of its short reasoning time and easy case learning.Combining case-based reasoning with BDI model can solve the drawback that case-based reasoning does not have autonomous learning.That CBR-BDI inference mechanism is applied to robots can make it have independent reasoning ability.In CBR-BDI reasoning,sentence similarity algorithm plays a very important role in case retrieval which is the first step of the reasoning.The accuracy of similarity algorithm is related to the reasonability.This dissertation introduces the importance of the sentence similarity algorithm in the CBR-BDI reasoning mechanism.By analyzing the advantages and disadvantages of the existing sentence similarity algorithm,a words semantic based sentence similarity computing method is proposed,which calculates the relative position offset of a combination of two words based on the central word of the sentence.And the relative positional offset of the words represents the structural information of the sentence and the word order information.Firstly,Sentence is preprocessed and the semantic similarity of words is calculated by Hownet.Central word of the sentence is determined by part of speech and grammatical rules,and the global qualifiers of the sentence are eliminated.Secondly,the relative position of words is marked according to the central word,and the relative position offset of the combination of two words is computed.At last,the information of sentence length difference,shallow structure information and semantic information are integrated to calculate sentence similarity.The experiment result proved that our proposed algorithm can effectively improve the comprehensive index of F-value.The improved sentence similarity algorithm is applied to the CBR-BDI inference system of the modular robot arm sorting system,which improves the intelligence of the sorting system and gives the user a better human-computer interaction experience.