Abstract

Vector Space Model (VSM) is widely used to represent documents and web pages. It is simple and easy to deal computationally, but it also oversimplifies a document into a vector, susceptible to noise, and cannot explicitly represent underlying topics of a document. A matrix representation of document is proposed in this paper: rows represent distinct terms and columns represent cohesive segments. The matrix model views a document as a set of segments, and each segment is a probability distribution over a limited number of latent topics which can be mapped to clustering structures. The latent topic extraction based on the matrix representation of documents is formulated as a constraint optimization problem in which each matrix (i.e., a document) A i is factorized into a common base determined by non-negative matrices L and R T, and a non-negative weight matrix Mi such that the sum of reconstruction error on all documents is minimized. Empirical evaluation demonstrates that it is feasible to use the matrix model for document clustering: (1) compared with vector representation, using matrix representation improves clustering quality consistently, and the proposed approach achieves a relative accuracy improvement up to 66% on the studied datasets; and (2) the proposed method outperforms baseline methods such as k-means and NMF, and complements the state-of-the-art methods like LDA and PLSI. Furthermore, the proposed matrix model allows more refined information retrieval at a segment level instead of at a document level, which enables the return of more relevant documents in information retrieval tasks.

Original languageEnglish (US)
Title of host publicationProceedings - IEEE International Conference on Data Mining, ICDM
Pages804-813
Number of pages10
DOIs
StatePublished - 2011
Event11th IEEE International Conference on Data Mining, ICDM 2011 - Vancouver, BC, Canada
Duration: Dec 11 2011Dec 14 2011

Other

Other11th IEEE International Conference on Data Mining, ICDM 2011
CountryCanada
CityVancouver, BC
Period12/11/1112/14/11

Fingerprint

Information retrieval
Vector spaces
Probability distributions
Websites

Keywords

  • Document clustering
  • Document representation
  • Matrix representation
  • Non-negative matrix approximation

ASJC Scopus subject areas

  • Engineering(all)

Cite this

Wang, X., Tang, J., & Liu, H. (2011). Document clustering via matrix representation. In Proceedings - IEEE International Conference on Data Mining, ICDM (pp. 804-813). [6137285] https://doi.org/10.1109/ICDM.2011.59

Document clustering via matrix representation. / Wang, Xufei; Tang, Jiliang; Liu, Huan.

Proceedings - IEEE International Conference on Data Mining, ICDM. 2011. p. 804-813 6137285.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Wang, X, Tang, J & Liu, H 2011, Document clustering via matrix representation. in Proceedings - IEEE International Conference on Data Mining, ICDM., 6137285, pp. 804-813, 11th IEEE International Conference on Data Mining, ICDM 2011, Vancouver, BC, Canada, 12/11/11. https://doi.org/10.1109/ICDM.2011.59
Wang X, Tang J, Liu H. Document clustering via matrix representation. In Proceedings - IEEE International Conference on Data Mining, ICDM. 2011. p. 804-813. 6137285 https://doi.org/10.1109/ICDM.2011.59
Wang, Xufei ; Tang, Jiliang ; Liu, Huan. / Document clustering via matrix representation. Proceedings - IEEE International Conference on Data Mining, ICDM. 2011. pp. 804-813
@inproceedings{c88fc1b8ed104c69bf00f461b6461534,
title = "Document clustering via matrix representation",
abstract = "Vector Space Model (VSM) is widely used to represent documents and web pages. It is simple and easy to deal computationally, but it also oversimplifies a document into a vector, susceptible to noise, and cannot explicitly represent underlying topics of a document. A matrix representation of document is proposed in this paper: rows represent distinct terms and columns represent cohesive segments. The matrix model views a document as a set of segments, and each segment is a probability distribution over a limited number of latent topics which can be mapped to clustering structures. The latent topic extraction based on the matrix representation of documents is formulated as a constraint optimization problem in which each matrix (i.e., a document) A i is factorized into a common base determined by non-negative matrices L and R T, and a non-negative weight matrix Mi such that the sum of reconstruction error on all documents is minimized. Empirical evaluation demonstrates that it is feasible to use the matrix model for document clustering: (1) compared with vector representation, using matrix representation improves clustering quality consistently, and the proposed approach achieves a relative accuracy improvement up to 66{\%} on the studied datasets; and (2) the proposed method outperforms baseline methods such as k-means and NMF, and complements the state-of-the-art methods like LDA and PLSI. Furthermore, the proposed matrix model allows more refined information retrieval at a segment level instead of at a document level, which enables the return of more relevant documents in information retrieval tasks.",
keywords = "Document clustering, Document representation, Matrix representation, Non-negative matrix approximation",
author = "Xufei Wang and Jiliang Tang and Huan Liu",
year = "2011",
doi = "10.1109/ICDM.2011.59",
language = "English (US)",
isbn = "9780769544083",
pages = "804--813",
booktitle = "Proceedings - IEEE International Conference on Data Mining, ICDM",

}

TY - GEN

T1 - Document clustering via matrix representation

AU - Wang, Xufei

AU - Tang, Jiliang

AU - Liu, Huan

PY - 2011

Y1 - 2011

N2 - Vector Space Model (VSM) is widely used to represent documents and web pages. It is simple and easy to deal computationally, but it also oversimplifies a document into a vector, susceptible to noise, and cannot explicitly represent underlying topics of a document. A matrix representation of document is proposed in this paper: rows represent distinct terms and columns represent cohesive segments. The matrix model views a document as a set of segments, and each segment is a probability distribution over a limited number of latent topics which can be mapped to clustering structures. The latent topic extraction based on the matrix representation of documents is formulated as a constraint optimization problem in which each matrix (i.e., a document) A i is factorized into a common base determined by non-negative matrices L and R T, and a non-negative weight matrix Mi such that the sum of reconstruction error on all documents is minimized. Empirical evaluation demonstrates that it is feasible to use the matrix model for document clustering: (1) compared with vector representation, using matrix representation improves clustering quality consistently, and the proposed approach achieves a relative accuracy improvement up to 66% on the studied datasets; and (2) the proposed method outperforms baseline methods such as k-means and NMF, and complements the state-of-the-art methods like LDA and PLSI. Furthermore, the proposed matrix model allows more refined information retrieval at a segment level instead of at a document level, which enables the return of more relevant documents in information retrieval tasks.

AB - Vector Space Model (VSM) is widely used to represent documents and web pages. It is simple and easy to deal computationally, but it also oversimplifies a document into a vector, susceptible to noise, and cannot explicitly represent underlying topics of a document. A matrix representation of document is proposed in this paper: rows represent distinct terms and columns represent cohesive segments. The matrix model views a document as a set of segments, and each segment is a probability distribution over a limited number of latent topics which can be mapped to clustering structures. The latent topic extraction based on the matrix representation of documents is formulated as a constraint optimization problem in which each matrix (i.e., a document) A i is factorized into a common base determined by non-negative matrices L and R T, and a non-negative weight matrix Mi such that the sum of reconstruction error on all documents is minimized. Empirical evaluation demonstrates that it is feasible to use the matrix model for document clustering: (1) compared with vector representation, using matrix representation improves clustering quality consistently, and the proposed approach achieves a relative accuracy improvement up to 66% on the studied datasets; and (2) the proposed method outperforms baseline methods such as k-means and NMF, and complements the state-of-the-art methods like LDA and PLSI. Furthermore, the proposed matrix model allows more refined information retrieval at a segment level instead of at a document level, which enables the return of more relevant documents in information retrieval tasks.

KW - Document clustering

KW - Document representation

KW - Matrix representation

KW - Non-negative matrix approximation

UR - http://www.scopus.com/inward/record.url?scp=84863121707&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84863121707&partnerID=8YFLogxK

U2 - 10.1109/ICDM.2011.59

DO - 10.1109/ICDM.2011.59

M3 - Conference contribution

AN - SCOPUS:84863121707

SN - 9780769544083

SP - 804

EP - 813

BT - Proceedings - IEEE International Conference on Data Mining, ICDM

ER -