Similarity Evaluation Techniques for Filtering Problems Vagan Terziyan
Similarity Evaluation Techniques for Filtering Problems ? Vagan Terziyan University of Jyvaskyla vagan@it. jyu. fi
Evaluating Distance between Various Domain Objects and Concepts - one of the basic abilities of an intelligent agent Are these two the same? … No ! The difference is equal to 0. 234
Contents 4 Goal 4 Basic Concepts 4 External Similarity Evaluation 4 An Example 4 Internal Similarity Evaluation 4 Conclusions
Reference Puuronen S. , Terziyan V. , A Similarity Evaluation Technique for Data Mining with an Ensemble of Classifiers, Classifiers In: A. M. Tjoa, R. R. Wagner and A. Al. Zobaidie (Eds. ), Proc. of the 11 th Intern. Workshop on Database and Expert Systems Applications, IEEE CS Press, Los Alamitos, California, 2000, pp. 1155 -1159. http: //dlib. computer. org/conferen/dexa/0680/pdf/06801155. pdf
Goal 4 The goal of this research is to develop simple similarity evaluation technique to be used for social filtering 4 Result of social filtering here is prediction of a customer’s evaluation of certain product based on known opinions about this product from other customers
Basic Concepts: Virtual Training Environment (VTE) 4 VTE is a quadruple: <D, C, S, P> • D is the set of goods D 1, D 2, . . . , Dn in the VTE; • C is the set of evaluation marks C 1, C 2, . . . , Cm , that are used to rank the products; • S is the set of customers S 1, S 2, . . . , Sr , who select evaluation marks to rank the products; • P is the set of semantic predicates that define relationships between D, C, S
Basic Concepts: Semantic Predicate P
Problem 1: Deriving External Similarity Values
External Similarity Values (ESV): binary relations DC, SC, and SD between the elements of (sub)sets of D and C; S and C; and S and D. ESV are based on total support among all the customers for voting for the appropriate connection (or refusal to vote)
Problem 2: Deriving Internal Similarity Values
Internal Similarity Values (ISV): binary relations between two subsets of D, two subsets of C and two subsets of S. ISV are based on total support among all the customers for voting for the appropriate connection (or refusal to vote)
Why we Need Similarity Values (or Distance Measure) ? 4 Distance between products is used to advertise the customers a new product based on evaluation of already known similar products 4 distance between evaluations is necessary to estimate evaluation error when necessary, e. g. in the case of adaptive filtering technologies used 4 distance between customers is useful to evaluate weights of all customers when necessary, e. g. to be able to integrate their opinions by weighted voting.
Deriving External Relation DC: How well evaluation fits the product Products Evaluation marks Customers
Deriving External Relation SC: Measures customer’s competence in the use of evaluation marks 4 The value of the relation (Sk, Cj) in a way represents the total support that the customer Sk obtains selecting (refusing to select) the mark Cj to evaluate all the products.
Example of SC Relation Evaluation marks Products Customers
Deriving External Relation SD: Measures customer’s competence in the products 4 The value of the relation (Sk, Di) represents the total support that the agent Sk receives selecting (or refusing to select) all the solutions to solve the problem Di.
Example of SD Relation Products Evaluation marks Customers
Normalizing External Relations to the Interval [0, 1] n is the number of products m is the number of evaluation marks r is the number of customers
Competence of a customer Evaluation marks Goods Conceptual pattern of evaluation marks definitions Di Conceptual pattern of goods’ features Competence in the goods Customer Competence in the evaluation marks Cj
Customer’s Evaluation: competence quality in Products
Customer’s Evaluation: competence quality in evaluation marks use
Quality Balance Theorem The evaluation of a customer’s competence (ranking, weighting, quality evaluation) does not depend on the competence area “virtual world of products” or “conceptual world of evaluation marks” because both competence values are always equal.
Proof . . .
An Example 4 Let us suppose that four customers have to evaluate three products from virtual shop using five different evaluation marks available. 4 The customers should define their selection of appropriate mark for every product. 4 The final goal is to obtain a cooperative evaluation result of all the customers concerning the quality of products.
C set (evaluation marks) in the Example Evaluation marks Nicely designed Expensive Easy to use Reliable Safe Notation C 1 C 2 C 3 C 4 C 5
S (customers) Set in the Example Customers IDs Fox Wolf Cat Hare Notation S 1 S 2 S 3 S 4
D (products) Set in the Example D 1 - Ultra Cast Spinning Reel D 2 - Nokia Communicator 9110 D 3 - i. Grafx Process Management Software
Evaluations Made for the Good “Reel” P(D, C, S) S 1 S 2 S 3 S 4 C 1 1 0+ 0 1 C 2 -1 -1** 0 -1 D 1 C 3 -1 0 ++ -1 0 C 4 0 1* 1 0 C 5 -1 -1*** 0 1 Customer Wolf prefers to select mark Reliable* to evaluate “Reel” and it refuses to select Expensive** or Safe***. Wolf does not use or refuse to use the Nicely designed+ or Easy to use++ marks for evaluation.
Evaluations Made for the Good “Communicator” P S 1 S 2 S 3 S 4 C 1 -1 1 1 -1 C 2 0 -1 -1 0 D 2 C 3 -1 -1 0 0 C 4 0 0 1 1 C 5 1 0
Evaluations Made for the Good “Software” P S 1 S 2 S 3 S 4 C 1 1 0 -1 -1 C 2 0 1 -1 -1 D 3 C 3 1 0 1 1 C 4 -1 -1 C 5 0 1 1 1
Example: Calculating Value DC 3, 4 D 3 P S 1 S 2 S 3 S 4 C 1 1 0 -1 -1 C 2 0 1 -1 -1 C 3 1 0 1 1 C 4 -1 -1 C 5 0 1 1 1
Resulting DC relation
Normalized and “Thresholded” DC relation 0 -1 0 0. 25 0. 5 1 0. 75 1
Result of Cooperative Goods Evaluation Based on DC Relation D 1 is nicely designed, reliable, not expensive, but not easy to use D 2 is reliable, safe, not expensive, but not easy to use D 3 is easy to use, safe, but not reliable
An Example: Calculating Value SD 1, 1
An Example: Calculating Value SC 4, 4
Resulting SD and SC relations
Normalized and “Thresholded” SD relation Fox Wolf Cat Hare Evaluations obtained from the customer Fox should be accepted if he evaluates goods similar to “Reels”. . . … or similar to “Software”. Fox’s evaluations should be rejected if they concern goods similar to “Communicator”
Normalized and “Thresholded” SD relation Fox Wolf Cat Hare Only evaluation from the customer Cat can be accepted if it concerns goods similar to “Communicator” All four customers are expected to give an acceptable evaluations concerning “Software” related goods
Normalized and “Thresholded” SC relation Nicely designed Expensive Easy to use Reliable Safe Fox Wolf Cat Hare Fox’s evaluations should be rejected if they concern design of goods Evaluation obtained from the customer Fox should be accepted if it concern usability (easy to use) of a good. . . … or reliability of a good.
Problem 2: Deriving Internal Similarity Values
Internal Similarity Values (ISV): binary relations between two subsets of D, two subsets of C and two subsets of S. ISV are based on total support among all the customers for voting for the appropriate connection (or refusal to vote)
Deriving Internal Similarity Values Via one intermediate set Via two intermediate sets
Internal Similarity for Customers: Goods-based Similarity Goods Customers
Internal Similarity for Customers: Evaluation marks-Based Similarity Evaluation marks Customers
Internal Similarity for Customers: Evaluation marks-Goods-Based Similarity Goods Evaluation marks Customers
Internal Similarity for Evaluation Marks Customers-based similarity Goods-customers-based similarity
Internal Similarity for Goods Customers-based similarity Evaluation marks-customers-based similarity
Normalized and “Thresholded” DDC relation similar neutral different
Conclusion 4 Discussion was given to methods of deriving the total support of each binary similarity relation. This can be used, for example, to derive the most supported goods evaluation and to rank the customers according to their competence 4 We also discussed relations between elements taken from the same set: goods, evaluation marks, or customers. This can be used, for example, to divide customers into groups of similar competence relatively to the goods evaluation environment
- Slides: 50