GB2378774A

GB2378774A - Searching procedures

Info

Publication number: GB2378774A
Application number: GB0110642A
Authority: GB
Inventors: James Richard Ling
Original assignee: One Stop To Ltd
Current assignee: One Stop To Ltd
Priority date: 2001-05-01
Filing date: 2001-05-01
Publication date: 2003-02-19
Also published as: GB0110642D0

Abstract

The invention provides a database searching procedure using a search processor, which is adapted to select a search methodology from a number thereof, and such a search processor. The procedure comprises the steps of receiving on the search processor a search request to retrieve specified data from a database; resolving the search request to generate criteria for the selection of the most appropriate search methodology; submitting the search request to a search engine which operates the selected methodology; and returning the results of the search from the search engine. The method and processor are particularly suited to computerised implementation and for searching on the Internet.

Description

Searching Procedures The present invention relates to a procedure for searching a database (as hereinafter defined) having a search engine associated therewith. The invention is particularly directed to a procedure for Internet and intranet searching. The present invention is also directed to a search processor for processing search requests.

The practice of electronic database searching is well known for both business and private use and various methods are used to conduct such searches. An increasingly common type of such searching is Internet searching where a user is generally required to access the web-site of a chosen search engine, enter a search request in the appropriate field or fields (often using the search engines'specific syntax) then to receive the results.

Search engines are associated with databases which contain information in a suitably sorted form. Search engines currently employ a variety of different search methodologies to address the database associated therewith. Examples of different methodologies currently available include conventional/traditional searching (also known as keyword, key-phrase, text or document searching), natural language searching (NLS), artificial intelligence searching (Al) and artificial logic searching (AL). Each of these has different advantages and drawbacks and is optimised for searching different types of information. Table 1 briefly summarises these differences.

Table 1

- Methodology Optimised for Database Field Conventional Text or Documents Description NLS Basic Product Product or Attribute Information Al or AL Advanced Product Product or Attribute Information and and metafile Sentence Structure

Conventional or traditional searching methodologies use keywords or key-phrases to select from a database items that contain the correct combination thereof. This has several drawbacks if words have a variety of constructions, derivatives and alternatives, but may be a powerful and targeted tooi in some circumstances. Natural language searching, has at least two advantages over conventional searching. Firstly, the search request may be posed in sentence style, by asking a question, rather than simply entering key words. The second is the way the database is searched for matching records. The NLS methodology searches the description, hierarchy and attributes fields for any given product rather than just the description. This is designed for product-based searches. This type of search can yield much more accurate information, by interrogating more appropriate descriptive fields associated with the item or range of items. The provision of a product catalogue is a key component in this method of searching.

Even though NLS has yet to be adopted widely, there is already a number of problems and shortcomings that have been identified. NLS was developed for a number of reasons, one of which was to address the information bias that conventional methodologies use. However, as a consequence NLS has a product-centric infrastructure. In the same way that

conventional methodologies were not suited to product-orientated material, NLS is not suited to information-orientated material.

The latest developments in search technology centre around AI and AL. These types of approach promise improvements in accuracy over NLS, particularly in semantic and lexical analysis as they should provide mechanisms to improve analysis of sentence structure, tense and meaning.

Despite these advantages, Al and AL still have inherent problems.

Their semantic or lexical analysis approach typically misinterprets meaning due to the difficulties associated with a computer's understanding of natural language, when rules of grammar and lexicons are applied explicitly to try to understand textual information. Parsing has an inherent inability to deal with ambiguity as it is based on a true/false decision-tree structure where one incorrect decision can jeopardise an entire analysis. Furthermore, the approach is language specific, and the reliance on the grammar of a given language is vulnerable to slang or grammatically incorrect constructions. A major drawback is that during initial configuration, the overheads and performance considerations required to incorporate the abstract concepts required for Al and AL into a search engine database combination, are extremely high. Finally, Al and AL are product-orientated like NLS and are not suited unsuited to basic document searches.

All existing search technologies are currently implemented in a mutually exclusive fashion and are not compatible with each other. In effect each subsequent new development is designed as a replacement for the methodologies that were used before. A database suited to one will function

poorly or not at all if addressed by a methodology different from the one for which it was designed.

The complexity and diversity of information available for example on the Internet alone is staggering and will only continue to grow in amount and variety. The users of the Internet have varied requirements and differing degrees of competence with the vast majority being inexperienced and unsure of how to find what they are looking for. This can lead to poor results and inefficient searching with recent statistics showing that 70% of people searching on the Internet fail to find what they are looking for. This is an unacceptabie degree of faiiure for people who rely on the sought for information being found. For most Internet users the complexity of searching not to mention the vast quantity of spurious results that can be generated by present means, often cause them to fail to locate the items for which they are searching or worse still to give up trying.

The present invention has stemmed from the appreciation that often different methodologies would be appropriate when searching for different types of information, but that a search request needs to be framed in appropriate/particular syntax or context to work effectively. However, only the most experienced of searchers are competent enough to select the appropriate methodology and phrase a search request in the correct terms, and this can still involve a degree of time consuming trial and error.

It is an aim of the present invention to provide a searcher (also referred to as a user) with the ability to conduct an efficient search specifically tailored to achieve optimum accuracy regardless of the information type or skill of the searcher. It is a further aim of the present invention to provide a way of

searching that is simple, adaptable and utilises an appropriate methodology to deliver accurate results.

Therefore, according to the present invention there is provided a database searching procedure using a search processor adapted to select a search methodology from a plurality thereof known to the search processor, which procedure comprises the steps of: - receiving on the search processor a search request to retrieve specified data from a database; - causing the search processor to resolve the search request to generate criteria for the selection of the most appropriate search methodology ; - submitting the search request to a search engine which operates the selected methodology on a database; and - returning the results of the search from the search engine.

The present invention is intended primarily to be used for electronic or computerised searching and therefore it is highly preferred that the procedure is carried out by a program or program steps running on computer apparatus.

The search request may be generated on a first computer with the procedure being carried out by a program or program steps running on the same computer. The search engine or engines and its/their associated database (s) could also be maintained in the same computer or could alternatively be located on remote computers connected by a data network.

The procedure could be carried out by a program or program steps running on a second computer in communication with the first, the search request being generated on the first computer and transmitted to the second.

This second computer could hold the search engine (s) and possibly also the database.

The procedure may also be performed with the search processor located on a third computer in the path of communication between a computer upon which the search request is generated and a computer or computers which operates the search engine or engines.

The search processor may direct the pre-processed search request to a variety of different search engines, each search engine operating a different methodology and having an associated database optimised therefor.

Alternatively, it is envisaged that some or aii of the different search engines may use a common database that is optimised for a number of different methodologies.

The present invention can select search methodologies including conventional keyword searching, natural language searching, artificial logic and artificial intelligence. It could also be simply adapted to incorporate any future search methodologies as and when they are developed.

The selection of the most appropriate methodology may be based on the selection criteria derived by the search processor. The derivation of the criteria may be based upon the phrasing, language, context wording or other parameters of the submitted search request and any specified data type.

Relevant factors could also include the required speed of transmission of data, prioritisation details, filtering details and the need for encryption.

In addition to the selection of the appropriate methodology, the search processor may perform some alteration on the search request. This preprocessing of the search will normally be to adapt it to the optimum style for

the selected methodology. For example, if a search request was originally posed in a partial sentence style, perhaps containing extraneous words such as"in","the","a", the search processor may determine that conventional searching is ideal. These words could be removed before the request is sent to the correct search engine.

Further alterations to the search request might include any of the following : . Removal of ambiguity if a search is likely to produce numerous results. a Simple spelling/grammar correction e. g. change the search request from"what is the bggest size denim shirt"to"largest denim shirt".

* Common mistake amendment Le. change "William Gates" to "Bill Gates". This could be done by reference to previous successful searches and also contextually (e. g. if the query also included the word"Microsoft").

* Streamline and focus the query. For example"Please show me the latest models of the Jaguar XK8 in Black with a V8 Engine"could be changed to"Jaguar XK8 in Black" # Re-context the query. For example in"Basic Language Interpretation" the normal focus would be on the spoken language such as English or Spanish. On closer examination the actual request is 99% more relevant to the Computer Programming Language. Therefore the context of the search request could be altered for programming language rather than natural language.

* Possible redirection of request.

According to a second aspect of the present invention there is also provided a search processor adapted to determine the optimum search methodology from a plurality thereof known to the search processor, which search processor comprises; - means for receiving a search request to retrieve specified data from a database; - means for resolving the search request to generate criteria for the selection of the most appropriate search methodology; and - means for submitting the search request to a search engine which operates the selected methodology on a database.

As described above, with reference to the method of the present invention, the search processor may be located in a number of positions. If, as is highly preferred, the search processor comprises a program or a plurality of program steps running on a computer, the search processor may be located on the computer of a person preparing a search request.

Alternatively it may be on a computer that operates the search engine or it could be on a computer in the path of communication therebetween.

The search processor may also be provided with means to alter the format of the search request in response to the selected methodology or generated criteria.

It is clearly important for both aspects of the present invention that a basic search request is provided. This search request can be delivered to the search processor is a any suitable way, such as by typing it into an appropriate field on a web-site and activating the system. An alternative means of submitting the search request is provided in our co-pending

International application No PCT/GB01/00703. In that invention a search request is submitted to a search engine as part of a longer string which also includes a locator for that search engine. The string is intercepted and parsed into the search request and the locator, with the locator being used to forward the search request as appropriate. If that invention were used in the context of the present invention, the locator would relate to the search processor, which would then communicate with the appropriate search engine. The search request parsed from the string originally provided would then be dealt with accordingly.

The results of a search may be sent directly back to the user by the search engine, or they may be handled on their return by the search processor. Therefore the search processor may be further provided with means to receive the results of the search from the search engine and return them to the originator of the search request. The results may be returned in their native form as received from the search engine, or instead may be manipulated or changed in some way. This may be in response to information derived from the original search request or otherwise specified by the user. Such manipulation could include prioritisation, filtering and sorting.

As used herein, the term"database"refers to the data source from which the search engine retrieves data to be returned to the user. Such a database could include, but is not limited to, conventional relational tables, lists of links or files, flat files etc. The term database also includes knowledge base and any other suitable information repository. Similarly the term search engine as used herein includes any program or computer with a predominate searching functionality, not just a conventional Internet search system.

In order that it may be better understood, and by way of example only, various embodiments of the present invention will now be described with reference to the accompanying drawings in which: Figure 1 is a simplified flow chart showing a first embodiment of the present invention operating with search engines accessing separate databases, and the search processor running on a server; Figure 2 is a simplified flow chart showing a second embodiment of the present invention operating with search engines accessing a common database, and the search processor running on a server; Figure 3 is a simplified flow chart showing a third embodiment of the present invention operating with search engines accessing separate databases, and the search processor running on a client computer; and Figure 4 is a simplified flow chart showing a fourth embodiment of the present invention operating with search engines accessing a common database, and the search processor running on a client computer.

Referring initially to Figure 1, there is shown a simplified flow chart representing an embodiment of the method of the present invention. The first stage of the process requires a user to enter a search request at step 10. Generally, this will be entered by typing an appropriate search request on the user's computer in an appropriate entry field. When the user has completed entry of the search request, the request is transmitted at step 11 to the search processor 12. In this embodiment, the transmission step is carried out over a suitable data network, such as the Internet, from a client computer of the user with the remainder of the process being carried out on a server or servers.

The search processor 12 receives the search request at step 13 and uses this to generate criteria for selection of a search methodology. The search processor 12 then uses this criteria to select the search methodology at step 14. Having selected the search methodology, the search processor 12 then forwards the search request to a search engine 16, 17, 18, 19 operating under the correct methodology. Methodologies currently employed include traditional key word searching 16, NLS 17, and Al/AL 18, but the present invention is able to incorporate any future methodologies (indicated in Figure 1 by the number 19). Once received, the selected search engine searches its associated database using the search request. Each search engine 16, 17, 18, 19 has a separate database 20, 21, 22, 23 optimized for the search methodologies particular requirements. As discussed elsewhere, the search request may be altered before transmission to the search engine to improve the searching.

The search engines operate in their normal fashion once an appropriate set of results has been generated and extracted from the database. The results must be returned to the original user for evaluation.

Native, or in effect raw, results may be returned directly to the user at step 24 from either the databases 20, 21, 22, 23 or the databases may return the results to the search engine 16, 17, 18, 19 which may then direct the results back to the user. Alternatively, the search processor 12 may receive the results from the search engines and subsequently forward them to the user at step 25. Before forwarding at step 25, the search processor may conduct some form of further processing of the results. Such further processing could

include prioritisation, sorting, categorisation, filtering and other forms of analysis of the data.

The procedure exemplified in Figure 2 is similar to that shown in Figure 1, however the search engines 16,17, 18,19 do not access separate independently-optimised databases. Instead, each searches the same database 28 which is specially designed to be addressable by all of the search engines.

In both Figures 1 and 2, the search processor, search engines and databases operate on a centralised server or servers. These are accessed as appropriate by a user from a client computer or work station. The client may communicate with the server across any suitable data communication network such as a LAN, WAN, or the Internet. The search request is transmitted in its raw form to the search processor and the final results are returned to the user over such a data network.

The embodiments of the present invention shown in Figures 3 and 4 are essentially similar to those depicted in Figures 1 and 2, respectively. The only difference is that instead of locating the search processor on a server, it is instead provided as part of the user's client computer. The client computer still communicates across a data network, but that communication involves the transmission of the preprocessed search request to the selected search engine, and the return of the raw results from either the search engine or the databases. The results can, as described above, be returned directly to the user or may be channelled through the search processor for further processing.

In the server-based examples in Figures 1 and 2, the search processor could form part of a particular search engine but is more likely to be provided separately so that it may direct the search request to appropriate search engines. The search processor might instead be located on a separate server in communication with the server carrying the appropriate search engines.

When the search processor is client-based, it could operate as an independent program on the client's computer which intercepts search requests generated by other programs. Preferably, the search processor will form an adjunct (or plug-in) to an existing Internet browser program.

The process by which the search request is resolved to generate criteria is usually active rather than passive, with the processor interpreting the input. The selection procedure is based on rules that are pre-specified to the processor, but which may be altered and updated as appropriate. Typically these will either be coded within the processor itself or will be available in a separate source such as a parameter file that a program reads before making a decision.

The rules may be based on a mixture of simple and complex criteria.

A highly simplified example could be: Is the SEARCH REQUEST in the form of a question? If NO then conventional searching may be the right route.

If YES then does the question contain complex structures? If YES then Al/AL may be employed.

If NO then NLS may be appropriate.

The above example is of course very simplified. In even the simples practical embodiment a decision is reached as to the type of methodology only after about twenty conditions or rules have been tested or applied. The rules provide a logical basis for the derivation of the criteria but will of course vary between implementation and are almost infinite in variation and type.

The rules could also be used to regulate possible advanced features of the invention. For example the search request may indicate that a secure connection is required if the nature of the results to be returned is sensitive such as financial records. Alternatively in some circumstances an enhanced speed of transmission may be desirable-in which case the criteria could initiate and control any compression technology that might be used.

The rules for resolving the search request are usually inter-related, hierarchical and sequentially arranged in so called "decision trees". Generally these "decision trees" will be large and complex, with for example a central high level tree, and then various subsets for rules and commands related to each methodology. Further sub-sub-sets might then control the content of the search request. Yet further sub-categories might exist governing more complex content or function such as statistical analysis.

Claims

Claims 1. A database searching procedure using a search processor adapted to select a search methodology from a plurality thereof known to the search processor, which procedure comprises the steps of : - receiving on the search processor a search request to retrieve specified data from a database; - causing the search processor to resolve the search request to generate criteria for the selection of the most appropriate search methodology ; - submitting the search request to a search engine which operates the selected methodology on a database; and - returning the results of the search from the search engine.
2. A database searching procedure as claimed in claim 1, wherein the procedure is carried out by a program or a series of program steps running on computer apparatus
3. A database searching procedure as claimed in claim 2, wherein the search request is generated on a first computer and the procedure is carried out by a program or series of program steps running on the same computer.
4. A database searching procedure as claimed in claim 2, wherein the search request is generated on a first computer and the procedure is carried out by a program or series of program steps running on a second computer in communication therewith.
5. A database searching procedure as claimed in claim 2, wherein the procedure is performed by a third computer in the path of communication

<Desc/Clms Page number 16>

between a first computer upon which the search request is generated and a second computer which operates the search engine.
6. A database searching procedure as claimed in any of the preceding claims, wherein a separate database optimised for a particular search methodology is searched when that particular methodology is selected.
7. A database searching procedure as claimed in any of claims 1 to 5, wherein a single database is searched and the database is optimised for searching using all the methodologies that the search processor may select.
8. A database searching procedure as claimed in any of the preceding claims wherein the search methodologies selectabie by the search processor include conventional keyword searching, natural language searching, artificial logic searching and artificial intelligence searching.
9. A database searching procedure as claimed in any of the preceding claims, wherein the criteria for selection of the most appropriate search methodology include at least one of the content to the submitted search request, the specified data, the format of the search request and the language of the search request.
10. A database searching procedure as claimed in any of the preceding claims, wherein the results of the search are returned to the search processor by the search engine and then returned to the user by the processor.
11. A database searching procedure as claimed in claim 10, wherein the data is sorted or filtered before being returned to the user.
12. A database searching procedure as claimed in any of the preceding claims, wherein an additional step of modifying the search request is carried out before submission to the search engine of that modified request.

<Desc/Clms Page number 17>
13. A database searching procedure as claimed in claim 13, wherein the modification alters the phrasing of the search request better to suit it to the selected search methodology.
14. A database searching procedure as claimed in claim 1 and substantially as herein described.
15. A search processor adapted to select the optimum search methodology from a plurality thereof known to the search processor, which search processor comprises ; - means for receiving a search request to retrieve specified data from a database; - means for resolving the search request to generate criteria for the selection of the most appropriate search methodology ; and - means for submitting the search request to a search engine which operates the selected methodology on a database.
16. A search processor as claimed in claim 15, which is further provided with means to receive the results of the search from the search engine and return them to the originator of the search request.
17. A search processor as claimed in claim 15 or claim 16, which processor forms part of the program steps running on a computer apparatus.
18. A search processor as claimed in clam 17, which is part of a first computer upon which the search request is entered.
19. A search processor as claimed in clam 17, which is part of a second computer upon which the search engine is provided.
20. A search processor as claimed in clam 17, which is part of a third computer which is in communication with a first computer upon which the

<Desc/Clms Page number 18>

search request is entered and a second computer upon which the search engine is provided.
21. A search processor as claimed in any of claims 15 to 20, in which a means for modifying the search request prior to submission to the search engine are further provided.
22. A search processor as claimed in claim 21, wherein the search request is modified to better suit the selected search methodology.
23. A search processor as claimed in claim 15 and substantially as herein described.