Computational intelligent analysis of lexical semantic networks: a multidimensional word sense association algorithm fusing fuzzy theory

Word sense association analysis in lexical semantic networks is a core task of natural language processing, but its fuzziness and multi-dimensional characteristics pose significant challenges to computational analysis. To address this problem, this paper constructs a computational intelligent analysis scheme integrating fuzzy theory. Firstly, word sense features are extracted from three dimensions (semantics, syntax, and pragmatics) to build a standardized multidimensional feature matrix. Secondly, the Fuzzy C-Means (FCM) clustering algorithm is adopted to mine potential semantic associations between words, and membership degree is used to quantify the fuzzy boundary of word senses. Finally, the Genetic Algorithm (GA) is introduced to optimize clustering parameters and improve analysis accuracy. Experiments are conducted on five million word-level Chinese corpora, and the results show that the algorithm achieves an accuracy of 87.6%, a recall of 83.5%, and an F1-score of 85.5%, which is significantly superior to traditional methods such as LDA and GloVe+K-Means. In application tests of information retrieval and machine translation scenarios, the scheme increases the F1-score to 68.6% and the translation accuracy to 85% respectively, verifying its practical value. This study enriches the theoretical methods of semantic network analysis and provides support for the upgrade of semantic understanding in natural language processing technology.