The present invention relates to a method for determining relevant advertisements to show to a search engine user (“User”) who has issued a search query against a search engine.
A search engine allows end-users to search for items within a data repository. Typically, a search engine user will provide to the search engine one or more search words (“Search Words”), and the search engine will show a list of pages that contain those words (“Search Results”), which may optionally contain an excerpt from each page (“Page Excerpt”) which may optionally show the Search Words in their context. One of the most common forms of a search engine is an Internet Search Engine (ISE), which maintains a list of words appearing in pages available over the Internet. Such search engines are well known in the art.
Search engines will often show a list of advertisements matching the Search Words. This is usually achieved by having the advertiser provide a list of words (“Keywords”), which, if matched by the Search Words, will cause the advertisement to appear. For example, an advertiser may choose the Keywords ‘volleyball’ and ‘shoes’, which will cause his advertisement to appear if the search engine user issues a query such as ‘best volleyball shoes for women’. This current method is limited in that the search engine user's objective may not always be easily predicted from his or her Search Words alone. Because of this, the search engine user may not be able to find a relevant advertiser to meet their needs and, similarly, the search engine may lose potential business in not presenting relevant advertisements that the search engine user might otherwise have clicked on. This current method also suffers because the process of generating all the possible Keywords by the advertiser can be cumbersome and incomplete, and, further, could result in outdated keyword lists.
The object of the invention is to provide more relevant advertisements to a search engine user and further to allow advertisements placed by advertisers to more easily be matched against a wider range of Search Words, without requiring advertisers to invest a substantial amount of effort in manually creating more Keywords. This object is achieved by providing a system and method for utilizing the Search Results, as opposed to utilizing the Search Words, as a means to generate richer information that can in turn be used to better match against relevant advertisements.
The search engine user issues a search query to a search engine. The search engine analyzes the search query and presents a list of relevant search results. The search engine then analyzes said search results and determines whether additional information is likely relevant to the search engine user's interest or objective in executing their search query. This additional information is fed into matching criteria that provides more relevant advertisements to the search engine user than they might have otherwise obtained had the search engine only used the Search Words to determine which advertisements were relevant.
According to the present invention there is provided a method of choosing advertisements to be shown to a search engine user, comprising the steps of: by an advertiser, providing an advertisement and at least one advertising criterion to a search engine; by a user, performing a search on the search engine using search words, the search providing search results; by the search engine, choosing an advertisement based on the search results and advertising criterion; and showing the chosen advertisement and the search results to the user.
According to the present invention there is provided a method of choosing advertisements to be shown to a search engine user, comprising the steps of: providing at least one advertising criterion; obtaining search results created by a search engine as a result of receiving and analyzing a search query made by the search engine user; analyzing the search results to produce analysis results; and using the analysis results to choose an advertisement matched to the at least one advertising criterion.
The present inventors have developed a method for choosing which advertisements to show to the User, based on the results of the search query he or she made, rather than on the Search Words alone. Since the User enters the Search Words for the purpose of receiving the Search Results, it is reasonable to assume that the Search Results are related to the objective of the User when making the query.
In step 200, an Advertiser (106) specifies an advertisement and corresponding Advertising Criteria for the advertisement
In step 202, a search engine User (100) submits Search Words to the search engine (102).
In step 204, the search engine uses the search index (103) to generate a set of Search Results for the User's request.
In step 206, the Search Results, optionally combined with the Search Words, are used to query the advertising database (104) to determine which advertisements to show based on the Advertising Criteria from step 200.
In step 208, the Search Results and the Advertisements are shown to the User.
A particularly inventive aspect of the present invention relates to step 206, in which the Search Results are used to query the advertisements database. One possible embodiment of Step 206 is described in more detail in a series of substeps 302-308, shown in
The Campaign Setup is the process by which an advertiser provides one criterion Advertising Criterion”) or more than one criterion (“Advertising Criteria”) for presenting his or her advertisement. The advertiser wishes to have the advertisement shown to the User whenever the Search Results match the Advertising Criterion or Advertising Criteria. One example of an Advertising Criterion is a word match. In this example, the advertiser provides a word or a set of words, and the advertisement appears if the word or set of words are found in the content of the documents referenced by the Search Results (the “Search Result Documents”) or in a corresponding Page Excerpt. A Page Excerpt is the small portion of text shown with the Search Results that highlights the search terms within the document.
It should be noted that all Advertising Criteria may be defined to match not only the Search Results shown to the User, but also Search Results that match the search query but are not shown to the User (e.g. a specific search may have 500 Search Results, but for simplicity the search engine will only show the first 10 to the User).
Another example of an Advertising Criterion is a website address or a Uniform Resource Identifier (URI). For example, an advertiser can enter a website address or URI of a corporate partner and request that the advertisement is shown whenever the website address or URI appears in the User's Search Results.
Yet another example of an Advertising Criterion is a meta-property. For example, an advertiser can enter a company name and request that the advertisement is shown whenever the User's Search Results contains a document whose properties specify that is was authored by said company.
The Search Results Analysis is the process in which the search engine determines which advertisements to show when the User makes a search query. The first step of Search Results Analysis is receiving the Search Results that were generated as a result of the search query. As mentioned, Page Excerpts, as well as the entire content of the Search Result Documents are considered part of the Search Results for the purpose of analysis. The second step of Search Results Analysis is to check which of the Advertising Criterion or Advertising Criteria previously provided by advertisers are related to the current Search Results. The third step of Search Results Analysis is advertisement prioritization. In some cases, the search engine may find that according to the Advertising Criteria it received for different advertisements, several advertisements should be presented to the User. In this case the search engine may want to omit some of them and/or present them in a certain order. The search engine may take into consideration such things as the type of each matching Advertising Criterion, its strength, and the amount of money the advertiser is willing to pay to have his advertisement used. Once the relevant advertisements are prioritized and filtered, the Search Results, as well as the prioritized list of advertisements are presented to the User.
One embodiment of the invention is implemented as follows and provided as Appendix I to this patent as working code for greater clarity. In this embodiment, we consider an Internet Search Engine (ISE, 102 in
1. The User executes a query through the web browser running on the PC against the ISE;
2. The ISE analyzes the search query and generates a list of relevant search results (202-204 in
3. The ISE then analyzes the search results (300-308 in
4. The ISE is queried for more advertisements by using the results from the analysis in step 3;
5. The search results, the original advertisement list from step 2, as well as the advertisements returned by the query in step 4 are presented to the User.
For example, if the Search Word is “BR50” and the Page Excerpt is “the BR50 Motorola battery provides your RAZR with three hours of nonstop . . . ”, the corresponding windows would contain:
For each window, all the word combinations are generated in step 304. For greater clarity, the use of the word “combination” is in the statistical sense, where we generate an exhaustive list of all word combinations and orderings. For example, if the window contained [a b c], this embodiment would generate:
Each unique word combination is then scored based upon its number of words and the number of times it appears in the Page Excerpts in step 306 as follows:
A threshold is determined in step 308, and all word combinations whose scores are less than this threshold are eliminated. Preferably, the threshold is given by the formula:
20 log2(number of Page Excerpts)
Although preferable, this formula is by no means limiting and other thresholds may be used to achieve the objects of the present invention. The remaining (over the threshold) word combinations will then contain words that are not one of the Search Words, but are contextually related to the Search Words (“Context Words”). These Context Words can then be used to re-query Advertisement Database 105. In this embodiment, the Context Words are fed back to the ISE with and without the original Search Words, in effect creating a larger search query. The ISE can then redisplay advertisements in the same fashion as it does today. Although preferable, this step, too, is not limiting and the Context Words may be fed back to the ISE with the original Search Words in various combinations to achieve the objects of the present invention.
Anyone with ordinary skill in the art will realize that this invention can be easily extended to include other forms of Advertising Criterion, which can use additional information from the Search Results, or additional information that can be derived from the Search Results. Several examples, which may be combined in many ways, include:
7. All Advertising Criteria above may be combined to produce a match score. This match score would be a function of whether different Advertising Criteria matched and the strength of such matches. Naturally, different weights may be given to different Advertising Criteria. In this case, a minimal match score may be provided as the Advertising Criterion.
The method of the present invention has the advantage that its Advertising Criteria are based on significantly more information (compared to just matching words in the User's search query). This allows showing advertisements that more accurately match the User's objective.
Another advantage of the present invention is that it allows an advertiser to use a much smaller list of Advertising Criteria. For example, today a retailer of Plasma TV sets will normally enter all known model codes of Plasma TVs as search words, in case a user searches for one of them (e.g. “W4200HD”, “FWD42PV1” and “TH42PV500”). This is a complicated, time consuming and error-prone process which can easily result in the generation of an outdated keyword list. In the present invention, requiring the words ‘Plasma’ and ‘TV’ to appear in the search results would most likely cover most such cases, since those words are likely to appear alongside the model number in the Page Excerpts (e.g. “The Dell W4200HD 42” High Definition Plasma TV delivers a wide aspect, high-definition resolution giving you life-like picture quality”).
While the invention has been described with respect to a limited number of embodiments, it will be appreciated that many variations, modifications and other applications of the invention may be made.
Appendix 1, attached as an integral part of this specification, presents commented source code written in PHP for a preferred embodiment of the invention
The present invention claims priority from U.S. Provisional Patent Application No. 60/817,451 dated Jun. 30, 2006, which is incorporated herein in its entirety and PCT/US2007/72620 filed Jul. 2, 2007.
Filing Document | Filing Date | Country | Kind | 371c Date |
---|---|---|---|---|
PCT/US2007/072620 | 7/2/2007 | WO | 00 | 8/28/2007 |
Number | Date | Country | |
---|---|---|---|
60817451 | Jun 2006 | US |