Web Usage Mining Personalization in Noisy, Dynamic, and Ambiguous Environments

0
0
1447 days ago, 570 views
PowerPoint PPT Presentation
Web Utilization Mining and Personalization in Loud, Dynamic, and Uncertain Situations. Olfa Nasraoui Learning Revelation and Web Mining Lab Dept of PC Designing and PC Sciences College of Louisville Email: olfa.nasraoui@louisville.edu URL: http://www.louisville.edu/~o0nasr01.

Presentation Transcript

Slide 1

Web Usage Mining & Personalization in Noisy, Dynamic, and Ambiguous Environments Olfa Nasraoui Knowledge Discovery & Web Mining Lab Dept of Computer Engineering & Computer Sciences University of Louisville E-mail: olfa.nasraoui@louisville.edu URL: http://www.louisville.edu/~o0nasr01 Supported by US National Science Foundation Career Award IIS-0133948 Nasraoui: Web Usage Mining & Personalization in Noisy, Dynamic, and Ambiguous Environments

Slide 2

Compressed Vita Endowed Chair of E-trade in the Department of Computer Engineering & Computer Science at the University of Louisville Director of the Knowledge Discovery and Web Mining Lab at the University of Louisville. Look into exercises incorporate Data Mining, Web mining, Web Personalization, and Computational Intelligence (Applications of transformative calculation and fluffy set hypothesis) . Filled in as program co-seat for a few meetings & workshops, including WebKDD 2004, 2005, and 2006 workshops on Web Mining and Web Usage Analysis, held in conjunction with ACM SIGKDD International Conferences on Knowledge Discovery and Data Mining (KDD). Beneficiary of US National Science Foundation CAREER Award. What I will talk about today is primarily the exploration items and lessons from a 5-year US National Science Foundation extend Nasraoui: Web Usage Mining & Personalization in Noisy, Dynamic, and Ambiguous Environments

Slide 3

My Collaborative Network? Nasraoui: Web Usage Mining & Personalization in Noisy, Dynamic, and Ambiguous Environments

Slide 4

Team: Knowledge Discovery & Web Mining Lab University of Louisville Director: Olfa Nasraoui (speaker) Current Student Researchers ( in order recorded ): Jeff Cerwinske, Nurcan Durak, Carlos Rojas, Esin Saka, Zhiyong Zhang, Leyla Zhuhadar Note: Gender adjusted & multicultural ;- ) Nasraoui: Web Usage Mining & Personalization in Noisy, Dynamic, and Ambiguous Environments

Slide 5

Past and Present Collaborators Raghu Krishnapuram, IBM Research Anupam Joshi, University of Maryland, Baltimore County Hichem Frigui, University of Louisville Hyoil Han, Drexel University Antonio Badia, University of Louisville Roberta Johnson, University Corporation for Atmospheric Research (UCAR) Fabio Gonzalez, Nacional University of Colombia Cesar Cardona, Magnify, Inc. Elizabeth Leon, Nacional University of Colombia Jonatan Gomez, Nacional University of Colombia Nasraoui: Web Usage Mining & Personalization in Noisy, Dynamic, and Ambiguous Environments

Slide 6

Introduction Information over-burden: a lot of data to filter/peruse through so as to discover fancied data Most data on Web is really superfluous to a specific client This is the thing that roused enthusiasm for procedures for Web personalization As they surf a site, clients leave an abundance of notable information about what pages they have seen, decisions they have made, and so on Web Usage Mining: A branch of Web Mining (itself a branch of information mining) that intends to find fascinating examples from Web use information (commonly Web Log information/clickstreams) (Yan et al. 1996, Cooley et al. 1997, Shahabi, 1997; Zaiane et al. 1998, Spiliopoulou & Faulstich, 1999, Nasraoui et al. 1999, Borges & Levene, 1999, Srivastava et al. 2000, Mobasher et al. 2000; Eirinaki & Vazirgiannis, 2003 ) Nasraoui: Web Usage Mining & Personalization in Noisy, Dynamic, and Ambiguous Environments

Slide 7

Introduction Web Personalization: Aims to adjust the Website as per the client's movement or interests (Perkowitz & Etzioni, 1997, Breeze et al. 1998, Pazzani, 1999, Schafer et al. 1999, Mulvenna, 2000; Mobasher et al. 2001, Burke. 2002, Joachims, 2002; Adomavicius &. Tuzhilin, 2005) Intelligent Web Personalization: regularly depends on Web Usage Mining (for client demonstrating) Recommender Systems: prescribe things important to the clients relying upon their advantage (Adomavicius & Tuzhilin, 2005) Content-based sifting: prescribe things like the things preferred by flow client (Balabanovic & Shoham, 1997) No thought of group of clients (practice just to one client) Collaborative separating: suggest things loved by "comparative" clients (Konstan et al., 1997; Sarwar et al., 1998; Schafer, 1999) Combine history of a group of clients: express (evaluations) or certain (clickstreams) Hybrids: join above (and others) Focus of our exploration Nasraoui: Web Usage Mining & Personalization in Noisy, Dynamic, and Ambiguous Environments

Slide 8

Some Challenges in WUM and Personalization Ambiguity : the level at which snaps are broke down ( URL A, B, or C as fundamental identifier) is extremely shallow, no significance Dynamic URLs : aimless URLs  significantly greater equivocalness Semantic Web Usage Mining: (Oberle et al., 2003) Scalability : Massive Web Log information that can't fit in primary memory requires procedures that are versatile (stream information mining) (Nasraoui et al.: WebKDD 2003, ICDM 2003) Handling Evolution : Usage information that progressions with time Mining & Validation in element situations: generally unexplored range… aside from in: (Mitchell et al. 1994; Widmer, 1996; Maloof & Michalski, 2000) In the Web utilization area: (Desikan & Srivastava, 2004; Nasraoui et al.: WebKDD 2003, ICDM 2003, KDD 2005, Computer Networks 2006, CIKM 2006) From Clicks to Concepts : couple of endeavors exist in light of relentless manual development of ideas, site metaphysics or scientific categorization How to do this naturally? (Berendt et al., 2002; Oberle et al., 2003; Dai & Mobasher, 2002; Eirinaki et al., 2003) Implementing recommender frameworks can be moderate, exorbitant and a jug neck particularly for scientists who need to perform tests on an assortment of sites For site proprietors that can't manage the cost of costly or confused arrangements Nasraoui: Web Usage Mining & Personalization in Noisy, Dynamic, and Ambiguous Environments

Slide 9

Different Steps Of our Web Personalization System STEP 1: OFFLINE PROFILE DISCOVERY STEP 2: ACTIVE RECOMMENDATION User profiles/User Model Post Processing/Derivation of User Profiles Site Files Recommendation Engine Preprocessing Recommendations Active Session Data Mining: Transaction Clustering Association Rule Discovery Pattern Discovery Server Logs User Sessions Nasraoui: Web Usage Mining & Personalization in Noisy, Dynamic, and Ambiguous Environments

Slide 10

Challenges & Questions in Web Usage Mining STEP 1: OFFLINE PROFILE DISCOVERY User profiles/User Model ACTIVE RECOMMENDATION Post Processing/Derivation of User Profiles Site Files Recommendation Engine Preprocessing Recommendations Active Session Data Mining: Transaction Clustering Association Rule Discovery Pattern Discovery Server Logs User Sessions Dealing with Ambiguity: Semantics? Understood scientific classification? (Nasraoui, Krishnapuram, Joshi. 1999) Website progression (can help disambiguation, yet constrained ) Explicit scientific classification? (Nasraoui, Soliman, Badia, 2005) From DB related w/dynamic URLs Content scientific categorization or cosmology (can help disambiguation, intense ) Concept chain of command speculation/URL pressure/idea deliberation : (Saka & Nasraoui, 2006) How does reflection influence nature of client models? Nasraoui: Web Usage Mining & Personalization in Noisy, Dynamic, and Ambiguous Environments

Slide 11

Challenges & Questions in Web Usage Mining STEP 1: OFFLINE PROFILE DISCOVERY User profiles/User Model ACTIVE RECOMMENDATION Post Processing/Derivation of User Profiles Site Files Recommendation Engine Preprocessing Recommendations Active Session Data Mining: Transaction Clustering Association Rule Discovery Pattern Discovery Server Logs User Sessions User Profile Post-handling Criteria? (Saka & Nasraoui, 2006) Aggregated profiles (recurrence normal)? Powerful profiles (rebate commotion information)? How would they truly perform ? How to approve? (Nasraoui & Goswami, SDM 2006) Nasraoui: Web Usage Mining & Personalization in Noisy, Dynamic, and Ambiguous Environments

Slide 12

Challenges & Questions in Web Usage Mining STEP 1: OFFLINE PROFILE DISCOVERY User profiles/User Model ACTIVE RECOMMENDATION Post Processing/Derivation of User Profiles Site Files Recommendation Engine Preprocessing Recommendations Active Session Data Mining: Transaction Clustering Association Rule Discovery Pattern Discovery Server Logs User Sessions Evolution: (Nasraoui, Cerwinske, Rojas, Gonzalez. CIKM 2006) Detecting & portraying profile advancement & change? Nasraoui: Web Usage Mining & Personalization in Noisy, Dynamic, and Ambiguous Environments

Slide 13

Challenges & Questions in Web Personalization STEP 1: OFFLINE PROFILE DISCOVERY User profiles/User Model ACTIVE RECOMMENDATION Post Processing/Derivation of User Profiles Site Files Recommendation Engine Preprocessing Recommendations Active Session Data Mining: Transaction Clustering Association Rule Discovery Pattern Discovery Server Logs User Sessions in the event of monstrous advancing information streams: Need stream information mining (Nasraoui et al. ICDM'03, WebKDD 2003) Need stream-based recommender frameworks ? (Nasraoui et al. CIKM 2006) How do stream-based recommender frameworks perform under development? How to approve above? (Nasraoui et al. CIKM 2006) Nasraoui: Web Usage Mining & Personalization in Noisy, Dynamic, and Ambiguous Environments

Slide 14

Challenges & Questions in Web Personalization STEP 1: OFFLINE PROFILE DISCOVERY User profiles/User Model ACTIVE RECOMMENDATION Post Processing/Derivation of User Profiles Site Files Recommendation Engine Preprocessing Recommendations Active Session Data Mining: Transaction Clustering Association Rule Discovery Pattern Discovery Server Logs User Sessions Implementing Recommender Systems: Fast , simple, adaptable , shoddy, free ? At any rate to help bolster inquire about… But Grand favorable position: help the little person…

SPONSORS