Understanding latent semantic indexing: A topological structure analysis using Q‐analysis

Li, Dandan; Kwong, C. K.

doi:10.1002/asi.21270

Cited by 11 publications

(10 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Kontostathis and Pottenger (2006) presented a mathematical proof that the SVD algorithm can encapsulate term cooccurrence information. Li and Kwong (2007a) studied VSM in general and LSI in particular from a completely new angle by exploring the topological structure of a corpus, in which Q-analysis devised by R. H. Atkin (1974) was applied to effect an analysis of the topological structure of the simplexes and their corresponding datasets. Kolda and O'Leary (1998) proposed to replace SVD with the semidiscrete decomposition (SDD).…”

Section: Literature Reviewmentioning

confidence: 99%

Unified linear subspace approach to semantic analysis

Kwong

Lee

2009

J. Am. Soc. Inf. Sci.

Self Cite

View full text Add to dashboard Cite

The Basic Vector Space Model (BVSM) is well known in information retrieval. Unfortunately, its retrieval effectiveness is limited because it is based on literal term matching. The Generalized Vector Space Model (GVSM) and Latent Semantic Indexing (LSI) are two prominent semantic retrieval methods, both of which assume there is some underlying latent semantic structure in a dataset that can be used to improve retrieval performance. However, while this structure may be derived from both the term space and the document space, GVSM exploits only the former and LSI the latter. In this article, the latent semantic structure of a dataset is examined from a dual perspective; namely, we consider the term space and the document space simultaneously. This new viewpoint has a natural connection to the notion of kernels. Specifically, a unified kernel function can be derived for a class of vector space models. The dual perspective provides a deeper understanding of the semantic space and makes transparent the geometrical meaning of the unified kernel function. New semantic analysis methods based on the unified kernel function are developed, which combine the advantages of LSI and GVSM. We also prove that the new methods are stable because although the selected rank of the truncated Singular Value Decomposition (SVD) is far from the optimum, the retrieval performance will not be degraded significantly. Experiments performed on standard test collections show that our methods are promising.

show abstract

Section: Literature Reviewmentioning

confidence: 99%

Unified linear subspace approach to semantic analysis

Kwong

Lee

2009

J. Am. Soc. Inf. Sci.

Self Cite

View full text Add to dashboard Cite

show abstract

“…Compared with the theories of graph and network, Q-analysis can explore n-ary relationships (i.e., the relationships among n entities) as well as the relationships between two things, which are more important for some practical problems [30] [13]. Q-analysis has been successfully applied to many fields, such as geographical studies [8] [31], database interaction [22], spatial databases [23] [14], and so forth.…”

Section: Related Workmentioning

confidence: 99%

A topology-based semantic location model for indoor applications

Lee

2008

Proceedings of the 16th ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems

Self Cite

View full text Add to dashboard Cite

Location-based services (LBSs) play more and more important roles in our daily life with the prevalence of mobile devices and the internet. Location modeling is a significant research topic in LBSs, which is needed to provide a well-defined representation of location knowledge for location browsing, navigation and query processing. In this paper, we propose that a topological structure can be attached to an exit-location space model, which can preserve the topology and distance semantics between locations (exits). The Q-analysis developed by R. H. Atkin is used to analyze the semantic information of the model. Compared with those existing models which only reveal the relationships between two entities, this novel model can provide the analysis of n-ary relationships (i.e., the relations among n entities) from both local and global viewpoints. Moreover, by using the rich structures obtained from the topological analysis, we define a semantic distance which can support more meaningful navigation and queries on complicated indoor environments. Examples are described in detail to demonstrate the effectiveness of our model.

show abstract

“…It is still considered powerful, thirty years later, in understanding complexity. In recent years, intense research efforts [5], [6], [7], [8], [9], [10], [11], [12], [13] argue for multidimensional networks to be the appropriate concept for the study of both the structure and the interactions of a wide range of complex systems.…”

Section: Introductionmentioning

confidence: 99%

“…40, no.1, pp. 71-77, 2007 9. Formal study of the "part-whole" relation instead of the traditional "is_a" hyperonimic relationship..…”

mentioning

confidence: 99%

A New Methodology for Collecting and Exploiting Vast Amounts of Dynamic Data

Legrand

Rousseaux

Soulier

et al. 2012

2012 Third International Conference on Emerging Intelligent Data and Web Technologies

View full text Add to dashboard Cite

The current approaches aiming at collective intelligence modelling often rely on traditional methods (ontologies, graphs). Even if those traditional methods may have reached their limitations in front of demanding emerging practices, the major conceptual tools enrolled for current and future Web are deeply rooted in the information storage and retrieval practices. The focus is on developing more original technologies for capturing, analyzing, exploiting and visualizing data. The agencements/arrangements provide the appropriate epistemological context of our contribution. The simplicial complexes are the mathematical support of the methodology. The result is a shift from networks studied towards graph theory to higher dimensional networks structures. The representation is more than graphs, or even hypergraphs. A geometric perspective shows the arrangement as assembling polyhedra of all sizes. Their contacts can form chains of adjacencies. It does not only generalize the notion of path graphs but it also makes available a range of quantitative and qualitative tools on the structure. Thus, separate parts, which can be more or less strongly linked, length of paths to traverse, and even loops or "missing parts", are meaningful metadata representations.

show abstract

Understanding latent semantic indexing: A topological structure analysis using Q‐analysis

Cited by 11 publications

References 31 publications

Unified linear subspace approach to semantic analysis

Unified linear subspace approach to semantic analysis

A topology-based semantic location model for indoor applications

A New Methodology for Collecting and Exploiting Vast Amounts of Dynamic Data

Contact Info

Product

Resources

About