Development of named entity recognition algorithm using long short-term memory and conditional random field as a hybrid technique

Name: Thiptanawat Phongwattana

Organization : Knowledge Discovery Company Limited. International A.I. Research Laboratory

Email : thiptanawat@kdlab.ai

keyword: Long short-term memory

ThaSH: Deep learning (Machine learning)

; Word Representation

ThaSH: Natural language processing (Computer science)

; Word embedding

ThaSH: Text data mining

; Global vector

; Named Entity Recognition

; Part-of-speech tagging

; Sentence tagging

Abstract: Currently, the data on the web has been dramatically increased and the popular definition of a kind of the data is "big data", however, the essence of data analysis is regarding deep data that the word is used to define useful information, practical use as well as it can be utilized. We found that over 90 percent of the data on the internet is a kind of text data. Hence, the key point of our research is challenging with information extraction from a ton of data from the internet that we call "text mining" in the terminology. To achieve this goal, we propose deep learning architecture. In general, a deep learning technique can be utilized in part-of-speech tagging to sentences, we found that long short-term memory (LSTM) recurrent neural nets (RNN) with a conventional named entity recognition (NER) method in text mining as a hybrid algorithm can optimize in classification and tagging of part-of-speech. With the state-of-the-art method, the limitation of LSTM can only make use of the neighboring tagging decisions. This research proposes a method to support LSTM making a decision of tagging words. The proposed method that we used is conditional random field (CRF) to outperform in NER tagging to sentences by making a decision of tags using other word embedding. The evaluation of the contextual word representation is also provided, which both accuracy and F1 score are over 90 percent at the end of 15 epochs.

King Mongkut's University of Technology North Bangkok. Central Library

Address: BANGKOK

Email: library@kmutnb.ac.th

Created: 2020

Modified: 2025-10-15

Issued: 2025-10-15

บทความ/Article

application/pdf

BibliograpyCitation : ใน มหาวิทยาลัยนเรศวร และสมาคมวิชาการทางวิศวกรรมไฟฟ้า (ประเทศไทย). การประชุมวิชาการทางวิศวกรรมไฟฟ้า ครั้งที่ 43 (EECON-43) (pp.390-393). พิษณุโลก : มหาวิทยาลัยนเรศวร

eng

©copyrights King Mongkut's University of Technology North Bangkok

ลำดับที่.	ชื่อแฟ้มข้อมูล	ขนาดแฟ้มข้อมูล	จำนวนเข้าถึง	วัน-เวลาเข้าถึงล่าสุด
1	EECON-43pp.390-393.pdf	1.2 MB

ใช้เวลา

0.016023 วินาที