Working Of

Search Engine

Guided By :- XxX

SOVAN MISRA , Roll no:-CS-07-42 , Regd no.- 0701230147

Why Engine?
Finding key information
from gigantic World Wide
Web is similar to find a
needle lost in haystack.
For this purpose we
would use a special
magnet that would
automatically, quickly and
effortlessly attract that
needle for us.
In this scenario magnet
is “Search Engine”
 What Is A Search
Engine..??

“Search Engine is a program that searches

documents for specified keywords and
returns a list of the documents where the
keywords were found.”
 Search Engine History

1990 - The first search engine Archie was released .

- The computers were interconnected by Telnet.

- File Transfer Protocol (FTP) used for

transferring files from computer to computer.

- Archie searched FTP servers and indexed their

files into a searchable directory.
1991 - Gopherspace came into existence with the
advent of Gopher.
Gopher cataloged FTP sites, and the resulting
catalog became known as Gopherspace .

1994 - WebCrawler, a new type of search engine

that indexed the entire content of a web page ,
was introduced.

1995 - Meta tags in the web page were first

utilized by some search engines to determine
relevancy.
1997 - Search engine rank-checking software was
introduced.

1998 - Search engine algorithms were being

introduced that increases the accurateness of
searching.

2000 - Marketers determined that pay-per click

campaigns were an easy yet expensive approach to
gaining top search rankings.
Eight reasonably well-known
Web search engines are : -
 Top 10 Search Providers by Searches, February 2010
Provider Searches (000) Share of Total
Searches (%)
All search 9,175,357 100.0

5,981,044 65.5
1,294,261 14.1
1,142,364 12.5
206,969 2.3
175,074 1.9
91,288 1.0
55,122 0.6
27,002 0.3

26,462 0.3
24,681 0.3

Others 110104 1.2

Source: Nielsen//Net Ratings, 2010

 Stages in information retrieval

Finding documents:
It is potentially needed to find required document
distributed over tens of thousands of servers.

Formulating queries:
It needed to express exactly what kind of
information is to retrieve.

Determining relevance:
The system must determine whether a document
contains the required information or not.
 Types of Search Engine

On the basis of working, Search engine is

categories in following group :-

Crawler-Based Search Engines

Directories
Hybrid Search Engines
Meta Search Engines
 I'm not as scary as I look...

Crawler-Based Search Engines

• It uses automated software programs to survey and

categories web pages , which is known as ‘spiders’,
‘crawlers’, ‘robots’ or ‘bots’.
• A spider will find a web page, download it and analyses the
information presented on the web page. The web page will
then be added to the search engine’s database.
• When a user performs a search, the search engine will check
its database of web pages for the key words the user
searched.
• The results (list of suggested links to go to), are listed on
pages by order of which is ‘closest’ (as defined by the ‘bots).
Examples of crawler-based search engines are:
• Google (www.google.com)
• Ask (www.ask.com)
 Robot Algorithm
All robots use the following algorithm for retrieving
documents from the Web:
1.The algorithm uses a list of known URLs. This list
contains at least one URL to start with.
2.A URL is taken from the list, and the corresponding
document is retrieved from the Web.
3.The document is parsed to retrieve information for the
index database and to extract the embedded links of
other documents.
4.The URLs of the links found in the document are added
to the list of known URLs.
5.If the list is empty or some limit is exceeded (number of
documents retrieved, size of the index database, time
elapsed since startup, etc.) the algorithm stops. otherwise
the algorithm continues at step 2.
Crawler program treats World Wide Web
as big graph having pages as nodes and
hyperlinks as arcs.
Crawler works with a simple goal:
Indexing all the keywords in web
pages’ titles.
Three data structure is needed for
crawler or robot algorithm
• A large linear array , url_table
• Heap
• Hash table
 Links Are Important

Where ever the

link I follow
Url_table :
• It is a large linear array that contains
millions of entries
• Each entry contains two pointers –
• Pointer to URL
• Pointer to title

Heap:
• It is a large unstructured chunk of virtual
memory where strings can be appended.
Hash table :
• It is third data structure of size ‘n’ entries.
• Any URL can be run through a hash
function to produce a nonnegative
integer less than ‘n’.
• All URL that hash to the value ‘k’ are
hooked together on a linked list.
• Every entry into url_table is also entered
into hash table.
• The main use of hash table is to start with a
URL and be able to quickly determine
whether it is already present in url_table.
 Data structure for crawler
Pointers Pointers
to URL to title Hash Overflow
chains
Code
2
String storage 0
URL 4
1
Title 5 19 6
2
21 44
URL
U 3
Title

Heap
n

Url_table Hash table

Building the index requires two phases :
Searching (URL proceesing )
Indexing.

The heart of the search engine is a

recursive procedure procees_url, which
takes a URL string as input.
Searching is done by procedure, procees_url as
follows :-

• It hashes the URL to see if it is already present in

url_table. If so, it is done and returns
immediately.
• If the URL is not already known, its page is
fetched.
• The URL and title are then copied to the heap
and pointers to these two strings are entered in
url_table.
• The URL is also entered into the hash table.
• Finally, process_url extracts all the hyperlinks
from the page and calls process_url once per
hyperlink, passing the hyperlink’s URL as the
 keyword Indexing
The stop list prevents indexing
of prepositions, conjunctions,
articles, and other words with
For each entry in url_table, indexing procedure will
many hits and little value.
examine the title and selects out all words not on
the stop list.
Each selected word is written on to a file with a line
consisting of the word followed by the current
url_table entry number.
when the whole table has been scanned , the file is
shorted by word.
 Formulating Queries

Keyword submission cause a request to be done in

the machine where the index is located(web server).

Then the keyword is looked up in the index

database to find the set of URL indices for each
keyword .

It is now indexed into url_table to find all the titles

and urls. Then it is stored in the Document server.

 These are then combined to form a web page and

sent back to user as the response.
 Determining Relevance

Classic algorithm "TF / IDF“ is used for

determining relevance.

 It is a weight often used in information retrieval

and text mining. This weight is a statistical
measure used to evaluate how important a word is
to a document in a collection

 A high weight in TF-IDF is reached by a high term

frequency (in the given document) and a low
document frequency of the term in the whole
collection of documents
 Term Frequency
• “Term Frequency” -The number of times a given term
appears in that document.
• It gives a measure of the importance of the term ti within
the particular document.

Term Frequency,

where ni is the number of occurrences of the considered

term, and the denominator is the number of occurrences of
all terms.
E.g.
If a document contains 100 total words and the word
computer appears 3 times, then the term frequency of the
word computer in the document is 0.03 (3/100)
 Inverse Document Frequency
The “inverse document frequency ”is a measure of
the general importance of the term (obtained by
dividing the number of all documents by the
number of documents containing the term, and then
taking the logarithm of that quotient).

Where,
• | D | : total number of documents in the corpus
•   : number of documents where the
term ti appears (that is ).
 Inverse Document Frequency
There are different ways of calculating the IDF

“Document Frequency” (DF) is to determine how

many documents contain the word and divide it by the
total number of documents in the collection.
E.g.
1) If the word computer appears in 1,000
documents out of a total of 10,000,000 then the
IDF is 0.0001 (1000/10,000,000).
2) Alternately, take the log of the document
frequency. The natural logarithm is commonly
used. In this example we would have
IDF = ln(1,000 / 10,000,000) =1/ 9.21
 Inverse Document Frequency
The final TF-IDF score is then calculated by
dividing the “Term Frequency” by the
“Document Frequency”.

E.g.
The TF-IDF score for computer in the
collection would be :
1)TF-IDF = 0.03/0.0001= 300 , by using first
formula of IDF.

2)If alternate formula used we would have

TF-IDF = 0.03 * 9.21 = 0.27.
The Big
Big Picture
Picture

You are here, and

make yourself
number one
You don't need to be #1 for everything.

You need to be #1 and that what

matters in case of searching.
Ranking procedure
 Directories

• The human editors comprehensively check the

website and rank it, based on the information
they find, using a pre-defined set of rules.

There are two major directories :

Yahoo Directory (www.yahoo.com)
Open Directory (www.dmoz.org)
 Hybrid Search Engines

Hybrid search engines use a combination of

both crawler-based results and directory
results.
Examples of hybrid search engines are:
Yahoo (www.yahoo.com)
Google (www.google.com)
 Meta Search Engines

• Also known as Multiple Search Engines or

Metacrawlers.
• Meta search engines query several other Web
search engine databases in parallel and then
combine the results in one list.
Examples of Meta search engines include:
Metacrawler (www.metacrawler.com)
Dogpile (www.dogpile.com)
This is how it works
(in short)
ANY QUERIES…..???
Hey!
This is sovan..
Please send your feedback @
[email protected]