SpanMTL: a span-based multi-table labeling for aspect-oriented fine-grained opinion extraction

Aspect-oriented Fine-grained Opinion Extraction (AFOE) aims to extract the aspect terms, corresponding opinion terms and sentiment polarity in a target sentence. Most previous methods treat AFOE as word-level or span-level task, which ignore complementarity of these two tasks. To integrate merits information, we construct an end-to-end Span-based Multi-Table Labeling (SpanMTL) framework. SpanMTL combines word-based span-based table labeling tackle task. Specifically, proposed model, use separate BiLSTMs encode information into 2D representation table. Based on table, with CNN by associating word-pair representations. At last, label distributions word- generate multi-table labeling. The method improves performances Pair (OPE) Triplet (OTE) tasks introducing span especially datasets lots spans. We have conducted various experiments validate our method. experimental results show that outperforms other baselines when sentences having information.