Personalization and Web-Mining

A big part of Web personalization as we experience it nowadays is based upon the core technologies of Web mining. In fact, in this chapter we have referenced a number of Web mining techniques (clustering, classification, association rules mining, sequential pattern discovery and prediction) when discussing the phase of data processing. Web mining is broadly defined as 'the use of data mining techniques for discovering and extracting information from Web documents and services and it is distinguished as Web content, structure or usage mining depending on which part of the Web is mined' (Kosala & Blockeel, 2000). It is a converging research area using techniques and methods that derive from various research fields such as: databases (DBs), information retrieval (IR), artificial intelligence (AI), as well as psychology and statistics.

The distinctions between the three main categories of Web-mining are not clear-cut . Web content mining might utilize text and links and even the profiles that are either inferred or explicitly inputted by users. The same is true for Web structure mining-it may use information about the links in addition to the link structures. Or, the traversed links can be inferred from the pages that were requested during user sessions and can be found recorded in server logs (which is a typical Web usage mining task). In practice, the three Web-mining categories can be used in isolation or combined in an application, especially in Web content and structure mining since links may be considered as part of the content of a Web document (Chakrabarti et al., 1999). In the majority of cases, Web applications base personalization on Web usage mining, which undertakes the task of gathering and extracting all data required for constructing and maintaining user profiles according to the 'logged' behavior of each user.

Web industry and researchers from diverse scientific areas have focused on various aspects of the topic. There are many research approaches and commercial tools that deliver personalized Web experiences based on business rules, Web site content and structure, as well as the user behavior monitoring. The most well-known applications of Web personalization at a research level include: Letizia (Lieberman, 1995), WebWatcher (Armstrong et al., 1995; Joachims et al., 1997), Fab (Balabanovic & Shoham, 1997), Humos/Wifs (Ambrosini et al., 1997), SiteHelper (Ngu & Wu, 1997), Personal WebWatcher (Mladenic, 1999), Let's Browse (Lieberman et al., 1999), SpeedTracer (Wu et al., 1998), WebPersonalizer (Mobasher et al., 2000), WebSIFT (Cooley et al., 1997, 1999b; Cooley et al., 2000), Web Utilization Miner - WUM (Spiliopoulou & Faulstich, 1998; Spiliopoulou et al., 1999a, 1999b; Spiliopoulou & Pohle, 2000), MIDAS (Buchner et al., 1999), IndexFinder (Perkowitz & Etzioni, 2000b).

Moreover, many vendors provide a variety of commercial tools that support mining for Web personalization. These tools can be integrated directly into a Web site server in order to provide users with personalized experiences.

• Net Perceptions (http://www.netperceptions.com): NetP 7.

• NetIQ Corporation (http://www. netiq .com): WebTrends Intelligence Suite, WebTrends Log Analyzer Series.

• Sane Solutions (http://www.sane.com): Funnel Web Analyzer, Funnel Web Profiler.

• Quest Software (http://www.quest.com): NetTracker 6.0 (Business Objects, Cognos, MicroStrategy).

• SAS (http://www.sas.com): SAS Value Chain Analytics, SAS IntelliVisor, Enterprise Miner.

• SPSS Inc. (http://www.spss.com): NetGenesis.

• WebSideStory, Inc. (http://www.websidestory.com): HitBox Services Suite (Enterprise, Commerce, Wireless Web Site Analysis).

• Accrue Software, Inc. (http://www.accrue.com): Accrue G2, Accrue Insight, Pilot Suite, Pilot Hit List.

• Blue Martini Software, Inc. (http://www.bluemartini.com): Blue Martini Marketing.

• Coremetrics, Inc. (http://www.coremetrics.com): Coremetrics Marketforce.

• E.piphany (http://www.epiphany.com): E.piphany E.6.

• Elytics, Inc. (http://www.elytics.com): Elytics Analysis Suite.

• IBM Corporation (http://www.ibm.com): WebSphere Personalization, SurfAid (Express, Analysis, Business Integration).

• Lumio Software (http://www.lumio.com): Re:cognition Product Suite, Re:action, Re:search, Re:collect.

• NCR Corporation (http://www.ncr.com): Teradata Warehouse.