Loading [MathJax]/extensions/MathMenu.js
A Framework for Semisupervised Feature Generation and Its Applications in Biomedical Literature Mining | IEEE Journals & Magazine | IEEE Xplore

A Framework for Semisupervised Feature Generation and Its Applications in Biomedical Literature Mining


Abstract:

Feature representation is essential to machine learning and text mining. In this paper, we present a feature coupling generalization (FCG) framework for generating new fe...Show More

Abstract:

Feature representation is essential to machine learning and text mining. In this paper, we present a feature coupling generalization (FCG) framework for generating new features from unlabeled data. It selects two special types of features, i.e., example-distinguishing features (EDFs) and class-distinguishing features (CDFs) from original feature set, and then generalizes EDFs into higher-level features based on their coupling degrees with CDFs in unlabeled data. The advantage is: EDFs with extreme sparsity in labeled data can be enriched by their co-occurrences with CDFs in unlabeled data so that the performance of these low-frequency features can be greatly boosted and new information from unlabeled can be incorporated. We apply this approach to three tasks in biomedical literature mining: gene named entity recognition (NER), protein-protein interaction extraction (PPIE), and text classification (TC) for gene ontology (GO) annotation. New features are generated from over 20 GB unlabeled PubMed abstracts. The experimental results on BioCreative 2, AIMED corpus, and TREC 2005 Genomics Track show that 1) FCG can utilize well the sparse features ignored by supervised learning. 2) It improves the performance of supervised baselines by 7.8 percent, 5.0 percent, and 5.8 percent, respectively, in the tree tasks. 3) Our methods achieve 89.1, 64.5 F-score, and 60.1 normalized utility on the three benchmark data sets.
Page(s): 294 - 307
Date of Publication: 30 September 2010

ISSN Information:

PubMed ID: 20876938

1 Introduction

With the exponential explosion of biomedical literature, such as MEDLINE, developing automatic text mining tools has become essential for people to seek information more accurately and efficiently. Biomedical text miming (BioTM) [12] becomes a hot area in data mining. The fundamental tasks, such as named entity recognition (NER), protein-protein interaction extraction (PPIE) and text classification (TC) have attracted a lot of research interests in various domains including Bioinformatics, natural language processing (NLP), and machine learning (ML). Although these tasks focus on extracting information of different formats, e.g., entities, relations or documents, classical methods usually treat them as the classification of text snippets and the methodologies have a lot in common. In traditional methods, each example is represented by a feature vector where each element is generated by a Boolean function indicating whether a word, n-gram or lexical pattern appears in the current example, and then these features are integrated in a supervised learning framework.

References

References is not available for this document.