LEADER 00000nam  2200325   4500 
001    AAI3290878 
005    20081006084558.5 
008    081006s2007    ||||||||||||||||| ||eng d 
020    9780549349655 
035    (UMI)AAI3290878 
040    UMI|cUMI 
100 1  Fullam, Karen Katherine 
245 10 Adaptive Trust Modeling in multi-agent systems: Utilizing 
       experience and reputation 
300    231 p 
500    Source: Dissertation Abstracts International, Volume: 68-
       12, Section: B, page: 8116 
500    Adviser: Suzanne Barber 
502    Thesis (Ph.D.)--The University of Texas at Austin, 2007 
520    Trust among individuals is essential for transactions. A 
       human or software agent in need of resources may reduce 
       transaction risk by modeling the trustworthiness of 
       potential partners. Experience- and reputation-based trust
       models have unique advantages and disadvantages depending 
       on environment factors, including availability of 
       experience opportunities, trustee trustworthiness dynamics,
       reputation accuracy, and reputation cost. This research 
       identifies how trusters may utilize both experience- and 
       reputation-based trust modeling to achieve more accurate 
       decision-making tools than using either modeling technique
       alone. The research produces: (1) the Adaptive Trust 
       Modeling technique for combining experience- vs. 
       reputation-based models to produce the most accurate 
       aggregated model possible, (2) a quantitative analysis of 
       the tradeoffs between experience- and reputation-based 
       models to determine conditions under which each type of 
       model is favorable, and (3) an Adaptive Cost Selection 
       algorithm for assessing the value of trust information 
       given acquisition costs. Experiments show that Adaptive 
       Trust Modeling yields an aggregate trust model more 
       accurate than either experience- or reputation-based 
       modeling alone, and Adaptive Cost Selection acquires the 
       optimal combination of trust information, maximizing a 
       truster's transaction payoff while minimizing trust 
       information costs. These tools enable humans and software 
       agents to make effective trust-based decisions given 
       dynamic system conditions 
590    School code: 0227 
590    DDC 
650  4 Artificial Intelligence 
650  4 Computer Science 
690    0800 
690    0984 
710 2  The University of Texas at Austin.|bElectrical and 
       Computer Engineering 
773 0  |tDissertation Abstracts International|g68-12B 
856 40 |uhttp://pqdd.sinica.edu.tw/twdaoapp/servlet/