Builing E-COMMERCE Applications And INFR

DATA WAREHOUSE PERSONALIZATION SYSTEMAN ANALYZE AND A COMPARATIVE STUDYKhalil CHARFIkhalilcharfi8@gmail.comAmir LADHARamir.ladhar@mail.comPresented bySupervised byM. Faiza GHOZZIAmir12IntroductionFirst ArticleSecond ArticleThird ArticleCOMPARATIVE STUDYConclusion

PRESENTATION OUTLINEamir2INTRODUCTION3Provide a quick access to relevant information.Using adapted interfaces to the user according to his preferences based on the concept of profile.

Personalization ??How ??amirPersonalization is a process of providing users with selected information on their specific needs.Personalization has been studied for many years and constitutes a hot topic in several domain like human computer interaction and databases, thus in the context of data warehouse the main idea of personalization is to ////This profile is used to model the user itself, his needs and his usual queries to provide more pertinent results.3INTRODUCTION4Typical problems in DW field:Large volumes of data.Burdening data exploration.While exploring previously unknown data, the OLAP query result may differ from expectations.Why personalization is important ?amir4

5First Article (OLAP Personalizaton with User-describing Profiles) ProblematicNatalija Kozmina and Laila Niedrite - Faculty of Computing, University of Latvia, Riga LV-1586, LatviaHow to provide user with the data he was looking for?amir1-Due to the large volumes of data the typical OLAP queries performed via OLAP operations by users may return too much information that sometimes makes further data exploration burdening or even impossible.5QuestionDescriptionProfile TypeWhat is the user expecting to get as a result ?User preferences dataPreferentialWho is the user ?Basic user data (personal data, session, activity, rights, etc.)UserWhere is the user located ?User physical location data & geolocation, according to user IP-addressSpatialWhen does the user interact with the system ?Time characteristics of user activities TemporalHow does the user & system interaction happen ?Characteristics of user device (i.e. PC, laptop, mobile phone, etc.), which is used for signing in as well as user software (e.g. web browser) characteristicsInteractionWhy the user is interested in this particular system ?User preferences are being gathered and analyzed. Recommendations are generated, according to user characteristics and preferences.Recommendational 6First Article (OLAP Personalization with User-describing Profiles) SolutionNatalija Kozmina and Laila Niedrite - Faculty of Computing, University of Latvia, Riga LV-1586, LatviaTo give a detailed characteristics of data warehouse user interaction with the system environment, Zachman Framework concept was used.User-describing profile diversitykhalilZachman Framework is an ontology that allows describing an arbitrary object from different viewpoints (temporary, spatial, etc. aspects).

67Refers to recommendational profile as recommendations are generated after analyzing of user preference profile.Are derivable from data warehouse log-tables.

In user profile, activity data indicates the intensity of usage of the reporting tool, defined by user hit count and spent time. Describes the environment, in which reporting tool is being used.

Context data are gathered automatically by means of web-services.

All the interaction profile attribute values are context data, as well as part of the spatial profile attributesAre gathered from data warehouse dimension attribute values.

All the temporal profile attributes values and part of spatial and user profile attribute values are static. User-describing profile data sources:Context data.Static data.Activity data.Analysis data.Explicitly entered data.First Article (OLAP Personalization with User-describing Profiles) SolutionNatalija Kozmina and Laila Niedrite - Faculty of Computing, University of Latvia, Riga LV-1586, Latvia

Data entered by user manually.

All the preferential profile values, which indicate the importance of user preference are gathered from the user explicitly. khalil78First Article (OLAP Personalization with User-describing Profiles) SolutionNatalija Kozmina and Laila Niedrite - Faculty of Computing, University of Latvia, Riga LV-1586, Latvia

khalilActivity diagram for recommendational profil

The user of data warehouse reporting tool signs in, the user-describing profiles are being displayed.

Each profile contains attributes with values, which are captured from data warehouse static, activity or context data sources.

The user is being informed with recommendations of the set of selected users.

If there are too many recommendations, the user is able to reduce its number by applying some filtering criterion.

When recommendations are filtered and the user has accepted the recommendation,

Then the observed set of user preferences is applied in the reporting tool.

As a result the user receives links to one ore more personalized reports.

89First Article (OLAP Personalization with User-describing Profiles) SolutionNatalija Kozmina and Laila Niedrite - Faculty of Computing, University of Latvia, Riga LV-1586, LatviaThe method, suggested in this paper, consists of the following steps: Stating questions (what? who? how? etc.) to describe the data warehouse user/system interaction.Identifying the user describing profiles.Collecting possible user-describing profiles attributes from various sources of information.Generating user characteristics via profile attributes after signing in the reporting tool.Suggesting possible recommendations for new and existing users of reporting tool, based on report preferences for the contents and structure of reports (OLAP preferences). Report personalization: applying selected recommendations to a report. khalilThe suggested method is composed of six steps to provide a personalized report to the user.

910Second Article (User Profile-Driven Data Warehouse Summary for Adaptive OLAP Queries) ProblematicNatalija Kozmina and Laila Niedrite - Faculty of Computing, University of Latvia, Riga LV-1586, LatviaHow to find relevant data in the data warehouses based on user preference ?Therefore, user profile change over time with user context changing.Hard preferenceSoft preference

AmirUser has two types of preferences:Hard preferences: Represent obligatory constraints of the user that must be firstly satisfied.Soft preferences: Represent optional constraints of the user.

1011Second Article (User Profile-Driven Data Warehouse Summary for Adaptive OLAP Queries) SolutionProviding personalized content to the user according to his profile preference to generate a data warehouse summary.This approach is based on three steps: Defining users profiles as a set of preferences.For each user profile, a data warehouse materialized view is created.Exploring this data warehouse materialized view during the user queries process.Natalija Kozmina and Laila Niedrite - Faculty of Computing, University of Latvia, Riga LV-1586, Latviakhalil11User Profile Definition12Second Article (User Profile-Driven Data Warehouse Summary for Adaptive OLAP Queries) SolutionNatalija Kozmina and Laila Niedrite - Faculty of Computing, University of Latvia, Riga LV-1586, LatviaDWUPV DB

Preferences InterfaceBeginnerExperienced UserQueryPersonalized AnswerOff-Line ModeUP DBUAUser profile-driven data warehouse materialized view architecture.One of the most important parts of data warehouse are its metadata

There are two metadata: User profile database (UP DB): Contains user preferences.User profile materialized view database (UPV DB): Contains materialized views generated by the system in off-line mode.

In order to identify decision makers, the user should go through authentication (UA).

If he is a beginner, the first time he connects to the system so he can enter his preferences.

Straight afterwards, the system records his profile in the user profile database (UP DB).

Then, the system creates the user-specific data warehouse view based on his profile and stores it in the UPV DB.1213Third Article (K-means Clustering Method for Recommending OLAP Queries) ProblematicData warehouses are characterized by voluminous data and are based on a user-centered analysis process.

Moreover, this data affect negativelyto the answers provided by the resultof OLAP.Natalija Kozmina and Laila Niedrite - Faculty of Computing, University of Latvia, Riga LV-1586, Latvia

AmirConsequently, how the authors will integrate personalization techniques within the OLAP process?

1314Third Article (K-means Clustering Method for Recommending OLAP Queries) SolutionNatalija Kozmina and Laila Niedrite - Faculty of Computing, University of Latvia, Riga LV-1586, Latvia

AMIRIn the framework, authors identify four main process:Knowledge acquisition.expilicit knowledge : User define his own knowledge !inmplicit knowledge : data minig technick to extract relevant clustues to represent significant and more elaborated olap queriesKnowledge integration into the data warehouse.Data warehouse schema evolution.OLAP queries personalization.

14COMPARATIVE STUDY151st Article2nd Article3rd ArticleType of work proposedInherited from Zachman framework conceptApproach and implementationApproach and implementationConcepts and technology usedZachman frameworkMaterialized viewUser preferencesRecommendation and Adaptation, Data mining, OLAP toolsData usedUser-describing profile dataData warehouse instanceImplicit knowledgeExplicit knowledgeKhalil15CONCLUSION16Personalization into the data warehousing process becomes a new trending research issue.

For future works it will be recommended to provide a personalized DW that integrate some of the presented concepts in one system : User preference & profile, Recommendation and Adaptationkhalil16REFERENCE17Natalija Kozmina, Laila Niedrite (2010). OLAP Personalization with User-Describing Profiles. Lecture Notes in Business Information Processing Volume 64, Pages 188 202.

Rym Khemiri and Fadila Bentayeb (2012). User Profile-Driven Data Warehouse Summary for Adaptive OLAP Queries. International Journal of Database Management Systems ( IJDMS ) Vol.4, No.6, Pages 69 84.

Fadila Bentayeb, Ccile Favre (2009). RoK: Roll-Up with the K-Means Clustering Method for Recommending OLAP Queries. Lecture Notes in Computer Science Volume 5690, Pages 501 515.

17THANK YOU FOR YOUR ATTENTION 18