[Short Text Classification of EMR Based on Entities and Dependency Parser]

Area of Science:

• Biomedical Informatics
• Natural Language Processing
• Data Mining

Background:

• Electronic Medical Records (EMR) are crucial for biomedical Big Data research.
• Effective text classification and mining of EMR are essential for advancing this research.

Purpose of the Study:

• To propose an improved method for classifying short texts within EMR data.
• To leverage entity dictionaries and dependency parsing for enhanced feature extraction.

Main Methods:

• Natural Language Processing (NLP) for text preprocessing: sentence segmentation, word segmentation, part-of-speech tagging, and entity extraction.
• Building entity dictionaries based on NLP results.
• Utilizing TF-IDF and Latent Semantic Analysis (LSA) for vocabulary feature selection.
• Employing dependency parsing to extract triple dependency relation features for classification.

Main Results:

• The proposed method, incorporating entity dictionaries and dependency parsing features, significantly improves classification performance compared to using vocabulary features alone.
• Experimental results show a notable increase in precision and F-value.

Conclusions:

• The integration of entity dictionaries and dependency parsing offers a powerful enhancement for EMR text classification.
• This method effectively improves the performance and accuracy of text mining in biomedical fields.