OntoSearch

Task Oriented Search is a web search executed with the purpose of assisting in the fulfillment of a specified task. Task Oriented Searches provide search results concerning different aspects of the task.

Before a search result can be provided to a Task Oriented Search, some preliminary processing is required by the Task Oriented Search Engine:

1. Task analysis and breakdown
2. Sub task definition, weight assignment and Sub Task Ranking
3. Sub task processing and ranking

When a search query is executed by a user, the search engine deducts according to keywords, natural language or other querying syntaxes, to which sub task the search is referring.  A sub task is then selected based on the Sub Task Ranking Matrix (STRM), and the search engine assigns the optimal, most probable super task. For the selected super task, the search engine will then produce a search result in the form of either textual links or mashup, according to the implementing search engine.

There are four types of Task Oriented Searches:

1. Task Oriented Information Query
2. Task Fulfillment Query (TFQ)
3.  Recursive Task Fulfillment Query (RTFQ)
4. Hybrid Task Oriented Search

See also:

1. Task Oriented Search Engine
2. Sub Task Ranking
3. Sub Task Ranking Matrix
4.Task Oriented Search Types

a. Task Oriented Information Query
b. Task Fulfillment Query – TFQ
c. Recursive Task Fulfillment Query – RTFQ
d. Hybrid Task Oriented Search

 

1. Task Oriented Search Engine

Task Oriented Search Engine is a niche search engine specializing in Task Oriented Searches, aimed to assisting users fulfill a certain task. Task Oriented Search Engine is one of the visions and more useful implementations of the semantic web (web3.0) where the internet becomes an information and context aware entity, and a Task Oriented Search Engine becomes an intelligent agent. With web3.0 springs sprouting towards 2010, the technology has yet to advance sufficiently for a fully computerized implementation of such search engine. One significant step in the direction of semantic search engines was done by Radar Networks’ Twine, semantically mapping a set of popular websites on the web. An entirely different approach was taken by Seartek.com, leveraging crowd sourcing and web2.0 to allow mapping of Task Oriented Searches by home web users. Task analysis and breakdown

 

2. Sub Task Ranking

In current day search engines, users are accustomed to breaking down their task to sub tasks before conducting numerous searches, for each subtask of their overall task. Since search behavior and habits is not likely to change, it is safe to assume users will NOT search for an overall task, but would conduct a search for a subtask of their overall task. It is then up to the search engine to deduct which overall task the user is attempting to accomplish.  The deduction of the super task is done using the Sub Task Ranking.

 

Sub task ranking is the assignment of grades for a super task depending on the probability of relevancy of the super task to the sub task in question.

Super-task and sub-task own an N to M multiplicity where a super-task can be composed of various sub-tasks, while a sub-task can be a component of various super-tasks. The modeling and grading of the super-task to sub-task relationship is conducted in a Sub Task Ranking Matrix, which is populated in the process of Sub Task Ranking.

While Task analysis and breakdown are performed top to bottom – super-task to sub-task, Sub Task Ranking is performed bottom to top, where a grade of a connection between a sub-task and a super-task is a grade of relevancy and probability of a super-task being the super-task of the sub-class as intended by the user performing the Task Oriented Search.

Sub Task Ranking in Task Oriented Search Engines can be executed in various different ways, from computerized data mining through crowd sourcing, and any combination of the two.

3. Sub Task Ranking Matrix – STRM

STRM is a matrix holding a set of grades of relevancy of super tasks to a set of sub tasks.

STRM models a complex N-M multiplicity where, as opposed to simple N-M multiplicity, each connection is assigned with a probability grade. Once populated, for each sub-task being searched, a set of super tasks will be fetched, and an optimal singular super task will be selected based on the probability grades assigned. Implementations of STRM vary and depend on the data modeling platform of the implementing search engines. Implementation can be done using a relational database such as Oracle and Microsoft SQL Server, or using a non relational approach such as Google’s Bigtable.

 

4. Task Oriented Search Types

a. Task Oriented Information Query
Task Oriented Information Query is a Task Oriented Search query performed in a web search engine, designed to retrieve information assisting in the fulfillment of a given task. Task Oriented Information Queries are commonly used in research where online information is invaluable tool in completing the task of preparing the scientific paper at hand.

b. Task Fulfillment Query – TFQ
TFQ is a Task Oriented Search Query performed in a web search engine, designed to load resources assisting in the fulfillment of a specified task. TFQ results will usually include websites allowing the user to perform actions, such as booking a flight, purchasing items online, submitting forms etc.

c. Recursive Task Fulfillment Query – RTFQ
RTFQ is a Task Oriented Search Query where the resulting task of the query is a sub task of a higher super task. When encountering RTFQ, the search engine may display the fetched sub task results, or perform a second iteration and fetch the super task of the fetched task. For example, a search query of “painting nursery walls” may produce a RTFQ where the super task is making a nursery, which may be a sub task of renovating your home. In such cases, the search engine may present results for making a nursery, or may choose to perform a second iteration and fetch results for renovating your home.
Determination of the behavior may depend on task substance and size, and on the search engine implementation.

d. Hybrid Task Oriented Search – Hybrid TOS
Hybrid Task Oriented Search is a Task Oriented Search comprised of any combination of different Task Oriented Query types. The most common and natural Hybrid TOS is a combination of Task Oriented Information Query and TFQ. In these cases, users are displayed results that both provide them information regarding a topic, as well as allow them to perform actions assisting in the completion of their task. A significantly less common and sometime problematic type of Hybrid TOS is a combination of RTFQ and any other search type. In these cases the decision of conducting a second iteration (and loosing valuable resources of the first iteration), or displaying results for the first iteration (and loosing valuable resources of the second iteration) is undetermined, and depends on different search engine implementations.