Scholarlink

<p>Named Entity Recognition (NER) is a significant task in natural language processing and plays a vital role in the field of e-commerce. NER identifies, categorizes and extracts the most important pieces of information from&#xa0;unstructured text&#xa0;without requiring time-consuming human analysis. A fusion attention mechanism with the pointer network is combined with Bidirectional LSTM (BiLSTM) and A Lite Bidirectional Encoder Representations from Transformers (ALBERT) model. The Conditional Random Field (CRF) is used in the proposed model to enhance attention on key elements while extracting entities and significant attributes. The ALBERT pre-trained model is combined to BiLSTM- CRF model for word embedding. The fusion attention mechanism is used for assigning the attentive weights for each word. With the fusion attention mechanism, pointer network for is included for generating pointer series to the input sequence elements. The performance of the proposed NER model is evaluated with the J&amp;L corpus and Kessler datasets. Simulation results show that the recognition accuracy is enhanced in better way than the previous NER methods.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

ALBERT-BiLSTM-CRF with pointer network based model for named entity recognition

  • 期刊论文

  • 发表日期：

    2025年7月26日

查看全文
  • J. Shobana,
  • K. M. UmaMaheswari

摘要

Named Entity Recognition (NER) is a significant task in natural language processing and plays a vital role in the field of e-commerce. NER identifies, categorizes and extracts the most important pieces of information from unstructured text without requiring time-consuming human analysis. A fusion attention mechanism with the pointer network is combined with Bidirectional LSTM (BiLSTM) and A Lite Bidirectional Encoder Representations from Transformers (ALBERT) model. The Conditional Random Field (CRF) is used in the proposed model to enhance attention on key elements while extracting entities and significant attributes. The ALBERT pre-trained model is combined to BiLSTM- CRF model for word embedding. The fusion attention mechanism is used for assigning the attentive weights for each word. With the fusion attention mechanism, pointer network for is included for generating pointer series to the input sequence elements. The performance of the proposed NER model is evaluated with the J&L corpus and Kessler datasets. Simulation results show that the recognition accuracy is enhanced in better way than the previous NER methods.