Unsupervised Feature Selection for Relation Extraction
نویسندگان
چکیده
This paper presents an unsupervised relation extraction algorithm, which induces relations between entity pairs by grouping them into a “natural” number of clusters based on the similarity of their contexts. Stability-based criterion is used to automatically estimate the number of clusters. For removing noisy feature words in clustering procedure, feature selection is conducted by optimizing a trace based criterion subject to some constraint in an unsupervised manner. After relation clustering procedure, we employ a discriminative category matching (DCM) to find typical and discriminative words to represent different relations. Experimental results show the effectiveness of our al-
منابع مشابه
کاهش ابعاد دادههای ابرطیفی به منظور افزایش جداییپذیری کلاسها و حفظ ساختار داده
Hyperspectral imaging with gathering hundreds spectral bands from the surface of the Earth allows us to separate materials with similar spectrum. Hyperspectral images can be used in many applications such as land chemical and physical parameter estimation, classification, target detection, unmixing, and so on. Among these applications, classification is especially interested. A hyperspectral im...
متن کاملAn Improved Discriminative Category Matching in Relation Identification
This paper describes an improved method for relation identification, which is the last step of unsupervised relation extraction. Similar entity pairs maybe grouped into the same cluster. It is also important to select a key word to describe the relation accurately. Therefore, an improved DF feature selection method is employed to rearrange low-frequency entity pairs’ features in order to get a ...
متن کاملUnsupervised Color Texture Feature Extraction and Selection for Soccer Image Segmentation
In this paper, we describe a new approach for color texture feature extraction and selection. We define color texture features as texture features which are computed by taking into account the color components of the pixels. We determine the most discriminating color texture features among a multidimensional set of color texture features by means of an iterative feature selection procedure asso...
متن کاملEmpirical Analysis of Supervised and Unsupervised Filter based Feature Selection Methods for Breast Cancer Classification from Digital Mammograms
In the design and development of an automated CAD tool for breast cancer detection and diagnosis, the various steps include enhancement, segmentation, feature extraction, feature selection and classification. The feature selection plays an important role in the design of the said CAD tool as it aims towards the redundant feature elimination and relevant feature selection. The selected feature s...
متن کاملDeep Understanding of Financial Knowledge through Unsupervised Learning
In this project, a universal information extraction method was implemented and applied to financial area, which supports aggregation and self analysis of complex information from massive correlated sources. In order to extract domain-independent relations between entities, open information extraction algorithm is used. Firstly, we actively label dataset using unsupervised learning algorithm by ...
متن کامل