Abstract:Correct and automatical semantic analysis has always been one of major goals in natural language understanding. However, due to the difficulties in deep semantic analysis, at present, the mainstream studies of semantic analysis are focused on semantic role labeling (SRL) and word sense disambiguation (WSD). Nowadays, these two issues are mostly considered as separate tasks. However, this approach ignores possible dependencies between them. In order to address the issue, an integrative semantic analysis model based on synergetic neural network (SNN) is proposed in this paper, which can easily express useful logic constraints between SRL and WSD. The semantic analysis process can be viewed as the competition process of semantic order parameters. The strongest order parameter will win by competition and desired semantic patterns will be recognized. There are three main innovations in this paper. First, an integrative semantic analysis model is proposed that jointly models word sense disambiguationand semantic role labeling. Second, integrative order parameter is reconstructed to reflect the relation among semantic patterns. Finally, integrative network parameters and integrative evolution equation are reconstructed, which can reflect the relationship of guiding and driving each other between word sense and semantic roles. The experiment results on OntoNotes 2.0 corpus shows the integrative method in this paper has a higher performance for semantic role labeling and word sense disambiguation, and provides a good practicability and a promising future for other natural language processing tasks.

Incorporating Contextual and Syntactic Structures Improves Semantic Similarity Modeling

Linguistic Modelling Based on Semantic Similarity Relation among Linguistic Labels

Neural language model and semantic compositionality model in semantic simi-larity

Bridging the Semantic Latent Space Between Brain and Machine: Similarity is All You Need

Exploiting Syntactic Structure for Better Language Modeling: A Syntactic Distance Approach

An Efficient Framework for Sentence Similarity Modeling.

Improving Text Understanding Via Deep Syntax-Semantics Communication

Better Combine Them Together! Integrating Syntactic Constituency and Dependency Representations for Semantic Role Labeling

Semantic Similarity Computing Model Based on Multi Model Fine-Grained Nonlinear Fusion

Learning Better Sentence Representation with Syntax Information

Syntax-Based Context Representation for Statistical Machine Translation.

Improved Semantic Representations From Tree-Structured Long Short-Term Memory Networks

Inference on Syntactic and Semantic Structures for Machine Comprehension.

Explicit Pairwise Word Interaction Modeling Improves Pretrained Transformers for English Semantic Similarity Tasks

Enhancing Semantic Representations of Bilingual Word Embeddings with Syntactic Dependencies

An Integration Model Of Semantic Annotation Based On Synergetic Neural Network

Improved Chinese Sentence Semantic Similarity Calculation Method Based on Multi-Feature Fusion

Semantic Similarity Analysis via Syntax Dependency Structure and Gate Recurrent Unit

Incorporating Complete Syntactical Knowledge for Spoken Language Understanding

Enhancing Unsupervised Semantic Parsing with Distributed Contextual Representations

Sentence Similarity on Structural Representations