Project Members: Nitu Singh Dnyaneshwari C External Guide: Amaresh Shirsat Internal Guide: Prof...

Post on 19-Dec-2015

218 views 2 download

Transcript of Project Members: Nitu Singh Dnyaneshwari C External Guide: Amaresh Shirsat Internal Guide: Prof...

RULES ENGINE FOR POPULARITY BASED RANKING

Project Members: Nitu Singh Dnyaneshwari C

External Guide: Amaresh Shirsat

Internal Guide: Prof Ketaki Naik

Sponsored By: Ubiqtas Technologies

AGENDA

Overview of Rules Engine Block diagram of Rules Engine Algorithms implemented Implementation details Merging rules and connector with the

framework Applications of rule engine Further enhancement of the project Questions please

OVERVIEW OF RULES ENGINE What is the rules engine? A central repository for ranking algorithms and various data sources.

What do we mean by rules?Algorithms which are used to rank data e.g page ranking algorithm.

Why are we using rule engine ?To generalize page ranking algorithm .

BLOCK DIAGRAM OF RULES ENGINE

Ranking algo

Source of data

connectors

Ranking algo & connector

User selects

Rules engine

Result

administrator

Designs the rules and appropriate connector and inserts them in framework

ALGORITHMS IMPLEMENTED

Web ranking -Page ranking algorithm -Keyword relevance -Connector to read web pages Database ranking -querying the database -searching of the data -ranking providing a recordset

PAGE RANKING ALGORITHM

What is a Page Ranking Algorithm? -A method to rank web pages. -Gives the page a numeric value. -Numeric value depends on the link

structure of the web.

continued…..

FACTORS GOVERNING A RANK OF THE PAGE

In links-Links pointing to a page

Out links-Links pointed by a page

Damping factor

A B C D

Dangling link Out-link from A

In-link to B

PR(C) = (1-dFactor) + dFactor PR (B) Out-links(B)Σ

Factors governing a rank of the page (contd…)

EXPANDED DEFINITION OF TERMS

PR(c):: PR stands for page rank of page c.

n=(1-d)::normalisation factor values range from 0 to 1.

d::damping factor value generally is 0.85 but it can range from 0 to 1.

EXPANDED DEFINITION OF TERMS (CONTD…)

PR(B)

Out-links(B)

The summation sign indicates sum of all the pages pointing to page C and divided by the number of out bound links of the page pointing to page C.

EXAMPLE

AT=

KEYWORD RELEVANCE ALGORITHM

Explanation of keyword relevance algorithm

The page which has maximum count of the

keyword is ranked high.

FACTORS GOVERNING OF KEYWORD RELEVANCE ALGORITHM Total count : total number of keywords

that occur in a webpage(repetition allowed)

Total keyword occurance :number of unique keywords that occur in a webpage(repetition not allowed)

EXAMPLE

if a page has keywords sun,moon,earth,moon then ,

total count = 4

Total keyword occurance=3

CONNECTOR

What is a Connector ?-Interface to read data from a specified data source.

Rules engine consist of two connectors -Web Connector -Database Connector

Web connector An application that gathers web graph

by navigating links

Data connector An application that gathers recordset

by navigating the travelling the record source

DIFFERENT TYPES OF CONNECTORS

IMPLEMENTATION DETAILS

PAGE RANKING ALGORITHM

Dynamic matrix of pages Using the formula for page ranking

Ranking of web pages

KEYWORD RELEVANCE ALGORITHM

Skimp through a web page

Removal of stopwords

Find out the keywords given by user in web page (repetitive and nonrepeatitive)

Rank according to total count and total keyword occurance

WEBCONNECTOR

Get the URL URL queue child URL linked list Scan child URL First URL in linked list acts as a parent URL

Go to that URL scan the entire page . URL’s present in the linked list are child

URL’s present in queue are parent.

QUERYING A DATABASE

Ms access is used at backend Different tables are used in order to

store the data set Normal pl-sql queries can be used to

query the the datbase e.g.Select *from person ;

RANKING AND PROVIDING A RECORDSET

It will perform searching through the database.

Use an sql query to rank the data

The connector collects data for which recommendation is required

The maximum recommended data is ranked high

MERGING RULES AND CONNECTOR WITH THE FRAMEWORK

Proj.dll

vc++ proj

C++

Windows application

Convert your vc project to com dll

Import your dll into the windows applic

Client side

dll

FURTHER APPLICATION OF THE PROJECT

SCRAPS=6055 FRIENDS=584

SCRAPS=2192 FRIENDS=203

SCRAPS=8872 FRIENDS=395

SCRAPS=401 FRIENDS=112

EXPLANATION OF E.G.

Design a rule for ranking a profile on basis of friends

Design a rule for ranking a profile on basis of scraps

Design a connector to get the information about number of friends and scraps

…CONTINUED

Suppose user selects rank on the basis of friends

Then in the above example “rahul” would be popped first then “sarang” then “shailu” and finally “dnyaneshwari”

REAL TIME APPLICATION OF THE PROJECT

QUESTIONS ?