Design of a Simple Intelligent Web Search
Agent
Amit Singh Dahal
G5638545 ITCS/M
 semantic gap between the user’s perception
of the search domain and the results
provided by search engines
 most current Internet search engines suffer
from Recall and Precision problems
 every search engine has an intelligent agent
 elements of intelligent agent, some
algorithms required to develop intelligent
Web Search Agent
1
 anything that gathers information or
performs some other service without your
immediate presence and on some regular
schedule
 perceiving its environment through sensors
and acting upon that environment with
effectors/actuators
2
3
 agent can generate all possible outcomes but
needs to filter from initial to current and
current to desired goal state
 used to find the desired goals by the user
efficiently and effectively
 two types of Algorithms
- Uninformed(blind search)
- Informed(heuristic search)
4
 No information of number of steps or the
path cost from the current state to goal state
 Searches include:
-depth first
-breadth first
-depth limiting
-iterative deepening
5
 have the information about the goal
 Information can be either estimated path cost
to the goal or estimated path cost away from
the goal
 Information known as “heuristic”
 Searches include:
-best first
-hill climbing
-A*(star)
6
 Web hunter should know:
-the target
-initial state/point
-number of iterations of target to find
-time frame(constraint)
-search methods
 For implementation:
-Socket programming
-HTTP/HTML
-Programming language
7
 Initialization:
-set up all variables, structures and arrays
-base information about ‘hunt’
-the target, initial state
-methods of searching
 Perception:
-using knowledge to contact and retrieve
the information
-identify if target is present and path to
other URL locations
8
 Action:
-takes all the information the system
knows and determines whether the goal
is met
-if met, the hunt is over else make
decision on where to go next
 Effect:
-list the location of target and give
feedback to users
9
10
 Testing the speed of searches and sorting
-time taken to search the target
-sorting the most common keywords first
 Using loops, algorithms and timing functions
 Searching the websites that has the maximum
number of hits by the user
11
 Searching with high speeds produce burden
to the servers
 Should be controlled using some timed waits
if Web Hunter make repeated queries to the
same servers
 During timed waits, the hunter can visit
different servers and speed can be
maintained
12
 Can integrate with Natural Language
Processing(NLP)
 Machine learning implementation to
understand users behavior
 In this way, improve the intelligence of Web
Based Agent
 Can be modified to Document Hunter or
Record Hunter
13
 Learned about:
-Web Hunting
-Role of Intelligent Web Search Agent
-Elements that need to be considered
when developing agent
 with the help of some information and tools,
Web Hunting can help the users to achieve
the goals in quick succession
 Keep in mind, “hunt safely”
14
THANK YOU!!!

15

Web hunting

  • 1.
    Design of aSimple Intelligent Web Search Agent Amit Singh Dahal G5638545 ITCS/M
  • 2.
     semantic gapbetween the user’s perception of the search domain and the results provided by search engines  most current Internet search engines suffer from Recall and Precision problems  every search engine has an intelligent agent  elements of intelligent agent, some algorithms required to develop intelligent Web Search Agent 1
  • 3.
     anything thatgathers information or performs some other service without your immediate presence and on some regular schedule  perceiving its environment through sensors and acting upon that environment with effectors/actuators 2
  • 4.
  • 5.
     agent cangenerate all possible outcomes but needs to filter from initial to current and current to desired goal state  used to find the desired goals by the user efficiently and effectively  two types of Algorithms - Uninformed(blind search) - Informed(heuristic search) 4
  • 6.
     No informationof number of steps or the path cost from the current state to goal state  Searches include: -depth first -breadth first -depth limiting -iterative deepening 5
  • 7.
     have theinformation about the goal  Information can be either estimated path cost to the goal or estimated path cost away from the goal  Information known as “heuristic”  Searches include: -best first -hill climbing -A*(star) 6
  • 8.
     Web huntershould know: -the target -initial state/point -number of iterations of target to find -time frame(constraint) -search methods  For implementation: -Socket programming -HTTP/HTML -Programming language 7
  • 9.
     Initialization: -set upall variables, structures and arrays -base information about ‘hunt’ -the target, initial state -methods of searching  Perception: -using knowledge to contact and retrieve the information -identify if target is present and path to other URL locations 8
  • 10.
     Action: -takes allthe information the system knows and determines whether the goal is met -if met, the hunt is over else make decision on where to go next  Effect: -list the location of target and give feedback to users 9
  • 11.
  • 12.
     Testing thespeed of searches and sorting -time taken to search the target -sorting the most common keywords first  Using loops, algorithms and timing functions  Searching the websites that has the maximum number of hits by the user 11
  • 13.
     Searching withhigh speeds produce burden to the servers  Should be controlled using some timed waits if Web Hunter make repeated queries to the same servers  During timed waits, the hunter can visit different servers and speed can be maintained 12
  • 14.
     Can integratewith Natural Language Processing(NLP)  Machine learning implementation to understand users behavior  In this way, improve the intelligence of Web Based Agent  Can be modified to Document Hunter or Record Hunter 13
  • 15.
     Learned about: -WebHunting -Role of Intelligent Web Search Agent -Elements that need to be considered when developing agent  with the help of some information and tools, Web Hunting can help the users to achieve the goals in quick succession  Keep in mind, “hunt safely” 14
  • 16.