Web Mining and Social Networking: Techniques and ... - tud.ttu.ee

146 7 Extracting and Analyzing Web Social NetworksWeb community slightly differs from a community of people, for example, a Web communitymay include competing companies. Since a Web community represents a certain topic, we canunderstand when and how the topic emerged and evolved in the Web.As introduced in Section 5.4, there are several algorithms for finding Web communities.Here, the extraction of Web community utilizes Web community chart that is a graph of communities,in which related communities are connected by weighted edges. The main advantageof the Web community chart is existence of relevance between communities. We can navigatethrough related communities, and locate evolution around a particular community.M.Toyoda and M. Kitsuregawa explain how Web communities evolve, and what kindsof metrics can measure degree of the evolution, such as growth rate and novelty. They firstexplain the details of changes of Web communities, and then introduce evolution metrics thatcan be used for finding patterns of evolution. Here the notations used are summarized in thissection.t 1 , t 2 , ..., t n : Time when each archive crawled. Currently, a month is used as the unit time.W(t k ): The Web archive at time t k .C(t k ): The Web community chart at time t k .c(t k ), d(t k ), e(t k ), ...: Communities in C(t k ).7.1.1 Types of ChangesEmerge: A community c(t k ) emerges in C(t k ), when c(t k ) shares no URLs with any communityin C(t k−1 ). Note that not all URLs in c(t k ) newly appear in W(t k ). Some URLs in c(t k )may be included in W (t k−1 ), and do not have enough connectivity to form a community.Dissolve: A community c(t k−1 ) in C(t k1 ) has dissolved, when c(t k−1 ) shares no URLs withany community in C(t k ). Note that not all URLs in c(t k−1 ) disappeared from W (t k−1 ). SomeURLs in c(t k−1 ) may still be included in W (t k ) losing connectivity to any community.Grow and shrink: When c(t k−1 ) in C(t k−1 ) shares URLs with only c(t k ) in C(t k ), and viceversa, only two changes can occur to c(t k−1 ). The community grows when new URLs areappeared in c(t k ), and shrinks when URLs disappeared from c(t k−1 ). When the number ofappeared URLs is greater than the number of disappeared URLs, it grows. In the reverse case,it shrinks.Split: A community c(t k−1 ) may split into some smaller communities. In this case, c(t k−1 )shares URLs with multiple communities in C(t k ). Split is caused by disconnections of URLsin SDG. Split communities may grow and shrink. They may also merge (see the next item)with other communities.Merge: When multiple communities (c(t k−1 )), d(t k−1 ), ...) share URLs with a single communitye(t k ), these communities are merged into e(t k ) by connections of their URLs in SDG.Merged community may grow and shrink. They may also split before merging.7.1.2 Evolution MetricsEvolution metrics measure how a particular community c(t k ) has evolved. For example, wecan know how much c(t k ) has grown, and how many URLs newly appeared in c(t k ). Theproposed metrics can be used for finding various patterns of evolution described above. Tomeasure changes of c(t k ), the community is identified at time t k−1 corresponding to c(t k ).This corresponding community, c(t k−1 ), is defined as the community that shares the mostURLs with c(t k ). If there were multiple communities that share the same number of URLs, acommunity that has the largest number of URLs is selected.