Ic service a service oriented approach to the development of recommendation systems
1 / 19

IC-Service: A Service-Oriented Approach to the Development of Recommendation Systems - PowerPoint PPT Presentation

  • Uploaded on

IC-Service: A Service-Oriented Approach to the Development of Recommendation Systems. Aliaksandr Birukou, Enrico Blanzieri, Vincenzo D'Andrea, Paolo Giorgini, Natallia Kokash, Alessio Modena. Introduction. Recommendation systems Service-Oriented Computing Implicit Culture

I am the owner, or an agent authorized to act on behalf of the owner, of the copyrighted work described.
Download Presentation

PowerPoint Slideshow about 'IC-Service: A Service-Oriented Approach to the Development of Recommendation Systems' - PamelaLan

An Image/Link below is provided (as is) to download presentation

Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author.While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server.

- - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript
Ic service a service oriented approach to the development of recommendation systems l.jpg

IC-Service: A Service-Oriented Approach to the Development of Recommendation Systems

Aliaksandr Birukou, Enrico Blanzieri, VincenzoD'Andrea, Paolo Giorgini, Natallia Kokash, Alessio Modena

ACM SAC, Seoul, Korea

Introduction l.jpg
Introduction of Recommendation Systems

  • Recommendation systems

  • Service-Oriented Computing

  • Implicit Culture

  • System for Implicit Culture Support (SICS)

  • SICS Architecture

    • Main modules

    • Configuration

  • Applications

    • Web service discovery

  • Conclusions

  • References

ACM SAC, Seoul, Korea

Recommendation systems l.jpg
Recommendation systems of Recommendation Systems

  • Prune large information spaces in searching for items of interest

  • Examples

    • movies (MovieLens),

    • music (JUKE-BOX),

    • books (Amazon),

    • hotels (TripAdvisor)

  • Meta-recommendation systems

    • Work with data from multiple (heterogeneous) information sources

    • MetaLens [Schafer et al., 2002]

ACM SAC, Seoul, Korea

Service oriented computing l.jpg

Service of Recommendation Systems









Service-oriented computing

Web service


  • Requirements for a recommendation service:

    • Use in various application domains

    • Ability to store heterogeneous client data

    • Adaptability to the needs of a particular client

    • Ability to process data according to the domain specific rules



Web service

ACM SAC, Seoul, Korea

Implicit culture ic motivation and goals l.jpg
Implicit Culture (IC): of Recommendation Systemsmotivation and goals

  • Communities of human/artificial agents have knowledge specific to their activities, i.e., community culture

  • The knowledge is often implicit and highly personalized

  • Encourage a newcomer to behave according to a community culture

  • Transfer knowledge implicitly (without special efforts for its analysis and description)

  • http://www.dit.unitn.it/~implicit

  • [Blanzieri et al., 2001]

ACM SAC, Seoul, Korea

Ic definitions l.jpg

Extract of Recommendation Systems

actions performed

in different situations


actions in

a given situation


agents’ actions

IC definitions

  • Action – something that can be done

  • Agent (actor) – somebody or something performing an action

  • Object – something that passively participate in the action

  • Situation – a state of the world faced by the agent. Includes a set of objects and a set of possible actions

  • Culture – a usual behavior of the group of agents

  • Group G – group of agents which behaviour is observed

  • Group G'– group of agents who require recommendations

  • Implicit Culture relation – situations in which agents of the group Gbehave similarly to agents of the group G'

  • System for Implicit Culture Support (SICS) – a system which tries to establish IC relation

ACM SAC, Seoul, Korea

System for implicit culture support sics l.jpg
System for Implicit Culture Support (SICS) of Recommendation Systems

Produce a theory about common user behavior

Produce recommendation about action

Stores information about actions

ACM SAC, Seoul, Korea

Sics architecture l.jpg
SICS Architecture of Recommendation Systems

  • SICS Core

    • SICS layer

      infers theory rules and recommends actions

    • Configuration and storage layer

      manages theory

  • SICS Remote Module

    defines protocols for information exchange with the client

  • SICS Remote Client

    provides a simple interface for remote clients

ACM SAC, Seoul, Korea

Storage module l.jpg
Storage Module of Recommendation Systems

  • Observations

    • Agents (1…N),

    • Actions (1),

    • Objects (0…N),

    • Attributes (0…N)

    • Scenes (1…N)

      • no agents

      • no timestamps

  • Theory rules

    • if consequent (predicates) then antecedent (predicates)

    • Predicates:

      • Conditions on observations (action- predicates)

      • Conditions on time (temporal-predicates)

ACM SAC, Seoul, Korea

Inductive module l.jpg
Inductive Module of Recommendation Systems

  • Analyses observations and generates theory rules for an actor or a group of actors

  • “Apriori” algorithm for mining association rules [Agrawal & Srikant, 1994]

    • A transaction is a sequence of executed actions A1,…,AN (can be obtained from observations using timestamps)

    • An association rule is an implicationof the form A1 A2 where A1, A2 are actions, A1 A2

    • The rule holds with confidencec if c% of transactions that contain A1 also contain A2

    • The rule A1 A2 has support s in the transaction set s% of transactions contain A1 A2

    • Generate association rules that have support and confidence greater than predefined minimum support and minimum confidence.

ACM SAC, Seoul, Korea

Composer module l.jpg
Composer Module of Recommendation Systems

  • Cultural Action Finder (CAF)

    • Matches actions executed by agents from group G’ with antecedents of the theory rules

      • Matching algorithms

    • Returns consequences of the theory rules (cultural actions)

  • Scene producer

    • Finds a set of agents that have performed actions similar to a cultural action for the agent X

    • Selects a set of agents similar to an agent X and a set of scenes S in which they have performed the actions

    • Select and propose to X a scene from S

ACM SAC, Seoul, Korea

Instance configuration l.jpg
Instance Configuration of Recommendation Systems

  • Composer constants:

    • Similarity threshold

    • Number of nearest neighbors

    • Return all scenes or only the best

    • Max number of observations

    • Names of groups G and G’

  • Configuration of similarity functions:

    • Rules for calculating similarity among observations

    • Similarity weights for elements (names and values)

      • exceptions, instants and default

    • Case sensitive or not

    • Regular expressions

  • Inductive Module constants

ACM SAC, Seoul, Korea

Applications l.jpg
Applications of Recommendation Systems

  • Prototypes:

    • Recommending Web links [Birukou et al., 2005]

    • Recommending scientific publications

  • Quality-based Indexing of Web Information (QUIEW) http://quiew.itc.it/

  • Supporting Polymerase Chain Reaction (PCR) experiments [Mullis et al., 1986] [Sarini et al., 2004]

  • Software patterns selection

  • Web service discovery

ACM SAC, Seoul, Korea

Web service ws discovery l.jpg
Web Service (WS) discovery of Recommendation Systems

  • Meeting functionality required by a user with specifications of existing web services

    • Problems: incomplete specifications, broken links, unfair providers…

  • Choosing a service with good quality characteristics

    • Problems: often QoS data are not available, some of them are context-dependent…

  • Implicit Culture approach

    • Analyze which web services have been previously used for similar problems by clients with similar interests

    • Use up-to-date information to improve service discovery and QoS-driven selection

ACM SAC, Seoul, Korea

A system for ws discovery l.jpg
A system for WS discovery of Recommendation Systems





ACM SAC, Seoul, Korea

Ws discovery in terms of ic l.jpg
WS discovery in terms of IC of Recommendation Systems

  • Observations

    • Actors

      • Applications (application name, user name, location)

      • Users (user name, location)

    • Objects

      • Operations (operation name, web service name)

      • Inputs/Outputs (parameter name, parameter value)

      • Requests (goals, operations, inputs/outputs)

    • Actions

      • Invoke (timestamp, operation, input)

      • Get response (timestamp, operation, output, response time)

      • Raise exception (timestamp, operation, exception type, input)

      • Provide feedback (timestamp, QoS parameters)

      • Submit request (timestamp, request)

  • Rules

    • if submit request(request) then invoke(operation-X(service-Y), request).

  • Similarity measures:

    • Vector Space Model (VSM)

      • Term Frequency- Inverse Document Frequency (TF-IDF) metric

    • WordNet-based semantic similarity measure

ACM SAC, Seoul, Korea

A system for ws discovery experimental results l.jpg

VSM of Recommendation Systems


A system for WS discovery: experimental results

  • 20 web services (http://www.xMethods.com) divided into 5 categories [Kokash et al., 2007]

  • 4 clients submit 100 requests

ACM SAC, Seoul, Korea

Conclusions l.jpg
Conclusions of Recommendation Systems

  • Ubiquity

    • The IC-service can be accessed from any workplace

  • Reusability

    • A unique solution for various distributed communities

  • Integration

    • The knowledge transfer between communities is facilitated

  • Scalability

    • 100000 observations of 100 users for one instance

    • Composition of several IC-Services is possible

  • Portability

    • XML storage

  • Customization

    • Ability of runtime configuring of theory rules…

ACM SAC, Seoul, Korea

References l.jpg
References of Recommendation Systems

  • [Schafer et al., 2002] J. B. Schafer, J. A. Konstan, and J. Riedl. Meta-recommendation systems: user-controlled integration of diverse recommendations. In Proc. of the Int. Conference on Information and Knowledge Management, pages 43-51. ACM Press, 2002.

  • [Blanzieri et al., 2001] E. Blanzieri, P. Giorgini, P. Massa, and S. Recla. Implicit culture for multi-agent interaction support. In CooplS: Proc. of the 9th Int. Conference on Cooperative Information Systems, volume 2172 of LNCS, pages 27-39. Springer, 2001.

  • [Birukov et al., 2005] A. Birukov, E. Blanzieri, and P. Giorgini. Implicit: An agent-based recommendation system for web search. In AAMAS: Proc. of the 4th Int. Joint Conference on Autonomous Agents and Multiagent Systems, pages 618-624. ACM Press, 2005.

  • [Mullis et al., 1986] K. B. Mullis, F. A. Faloona, S. Scharf, R. K. Saiki, G. Horn, H. A. Erlich. Specific enzymatic amplification of DNA in vitro: the polymerase chain reaction. In Cold Spring Harbor Symposia on Quantitative Biology, volume 51, pages 263-273, 1986.

  • [Sarini et al., 2004] M. Sarini, E. Blanzieri, P. Giorgini, C. Moser. From actions to suggestions: supporting the work of biologists through laboratory notebooks. In COOP: Proc. of 6th Int. Conference on the Design of Cooperative Systems, pages 131-146. IOS Press, 2004.

  • [Agrawal & Srikant, 1994] R. Agrawal and R. Srikant. Fast algorithms for mining association rules in large databases. In VLDB: Proc. of the 20th Int. Conference on Very Large Data Bases, pages 487-499. Morgan Kaufmann, 1994.

  • [Kokash et al., 2007] N. Kokash, A. Birukou, V. D'Andrea: Web service discovery based on past user experience. In: International Conference on Business Information Systems (BIS), to appear, Springer (2007)

ACM SAC, Seoul, Korea