Text Mining in Social Networks
Social networks are rich in various kinds of contents such as text and multimedia. The ability to apply text mining algorithms effectively in the context of text data is critical for a wide variety of applications. Social networks require text mining algo
- PDF / 493,703 Bytes
- 26 Pages / 439.37 x 666.142 pts Page_size
- 99 Downloads / 244 Views
Haixun Wang Microsoft Research Asia Beijing, China 100190 [email protected]
Abstract
Social networks are rich in various kinds of contents such as text and multimedia. The ability to apply text mining algorithms effectively in the context of text data is critical for a wide variety of applications. Social networks require text mining algorithms for a wide variety of applications such as keyword search, classi¿cation, and clustering. While search and classi¿cation are well known applications for a wide variety of scenarios, social networks have a much richer structure both in terms of text and links. Much of the work in the area uses either purely the text content or purely the linkage structure. However, many recent algorithms use a combination of linkage and content information for mining purposes. In many cases, it turns out that the use of a combination of linkage and content information provides much more effective results than a system which is based purely on either of the two. This paper provides a survey of such algorithms, and the advantages observed by using such algorithms in different scenarios. We also present avenues for future research in this area.
Keywords:
Text Mining, Social Networks
C. C. Aggarwal (ed.), Social Network Data Analytics, DOI 10.1007/978-1-4419-8462-3_13, © Springer Science+Business Media, LLC 2011
354
1.
SOCIAL NETWORK DATA ANALYTICS
Introduction
Social networks are typically rich in text, because of a wide variety of methods by which users can contribute text content to the network. For example, typical social networks such as Facebook allow the creation of various text content such as wall posts, comments, and links to blog and web pages. Emails between different users can also be expressed as social networks, which can be mined for a variety of applications. For example, the well known Enron email database is often used in order to mine interesting connections between the different characters in the underlying database. Using interesting linkages within email and newsgroup databases in addition to the content [5, 8] often leads to qualitatively more effective results. Social networks are rich in text, and therefore it is useful to design text mining tools for a wide variety of applications. While a variety of search and mining algorithms have been developed in the literature for text applications, social networks provide a special challenge, because the linkage structure provides guidance for mining in a variety of applications. Some examples of applications in which such guidance is available are as follows: Keyword Search: In the problem of keyword search, we specify a set of keywords, which are used to determine social network nodes which are relevant to a given query. In the problem of keyword search, we use both the content and the linkage behavior in order to perform the search. The broad idea is that text documents containing similar keywords are often linked together. Therefore, it is often useful to determine closely connected clusters of nodes in the soc
Data Loading...
