Abstract
Named entity recognition (NER) is an indispensable and very important part of many natural language processing technologies, such as information extraction, information retrieval, and intelligent Q & A. This paper describes the development of the AL-CRF model, which is a NER approach based on active learning (AL). The algorithmic sequence of the processes performed by the AL-CRF model is the following: first, the samples are clustered using the k-means approach. Then, stratified sampling is performed on the produced clusters in order to obtain initial samples, which are used to train the basic conditional random field (CRF) classifier. The next step includes the initiation of the selection process which uses the criterion of entropy. More specifically, samples having the highest entropy values are added to the training set. Afterwards, the learning process is repeated, and the CRF classifier is retrained based on the obtained training set. The learning and the selection process of the AL is running iteratively until the harmonic mean F stabilizes and the final NER model is obtained. Several NER experiments are performed on legislative and medical cases in order to validate the AL-CRF performance. The testing data include Chinese judicial documents and Chinese electronic medical records (EMRs). Testing indicates that our proposed algorithm has better recognition accuracy and recall rate compared to the conventional CRF model. Moreover, the main advantage of our approach is that it requires fewer manually labelled training samples, and at the same time, it is more effective. This can result in a more cost effective and more reliable process.
Highlights
With the continuous popularization of Internet and mobile Internet and the continuous improvement of information infrastructure in various domains, the available digital resources have grown explosively in our metaindustrial societies [1]
We propose a new approach called active learning (AL)-conditional random field (CRF), aiming to improve the efficiency recognition of the CRF model, and to decrease the number of annotated training samples as well. e testing experiments on medical and legislative fields have proven that our proposed method can produce a more efficient Named entity recognition (NER) model with fewer training samples, which can effectively cut the cost of manual annotation and improve the overall efficiency
Statistical methods mainly include hidden Markov models (HMM) [11], maximum entropy (ME) [12], support vector machines (SVM) [13], and CRF [14]. ese kinds of methods use the labelled corpus data to train the model combined with statistical probability. ey are transplantable and they have comparatively short construction periods they have more strict requirements for feature selection and much more dependence on the corpus. e mixed methods model is a combination of rules, dictionary-based approaches, and statistical ones, and they combine the advantages of both. ey employ rules to filter the target text in advance, and they are reducing the state of the search space, based on statistical methods
Summary
With the continuous popularization of Internet and mobile Internet and the continuous improvement of information infrastructure in various domains, the available digital resources have grown explosively in our metaindustrial societies [1]. NER is a process of recognizing and classifying words or phrases with special characteristics or meanings in a text. It belongs to the category of unsigned word recognition in lexical analysis, and it is an indispensable part of information extraction and retrieval, intelligent Q & A, and other natural language processing technologies [4]. E most commonly used relative approaches can be divided into three categories: rule and dictionary-based methods, statistical, and mixed ones. NER mainly used rule and dictionary-based methodologies, which require the design and development of the rule sets by domain experts and the use of proper linguists. E mixed methods model is a combination of rules, dictionary-based approaches, and statistical ones, and they combine the advantages of both.
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
