Patent Application
20080235204
Not applicable.
Not applicable.
Products and services evolve and improve over time through the comments, suggestions, and feedback of the general public. The general public can be a good source of feedback as it is the public that uses these products and services on a regular basis, and the public may be able to observe most flaws within the product or service that may need to be corrected.
There are many searching services available to the public that can search for a wide-range of information for a requesting user. Many of these searching services are based on different algorithms that are used to find the most relevant search results for a user's search request. However, these algorithms are not always perfect and may not always provide the best search results for a user's search request. Additionally, a search result set for a particular search query may be adequate for one person, however, the same search result for the same search query may not be sufficient for another person because everyone's needs and likes are not the same. Searching services, therefore, could be another service that could benefit from being subjected to user comments and feedback. Incorporating user feedback into the searching services' method of providing search results could lead to better and more relevant search results as the public's own feedback can be used to offer a better sense of what the public wants instead of just relying on algorithms to provide results.
A system and method for managing feedback data to be used for ranking search results is presented. The system and method discloses aggregating elements of feedback data from a plurality of users, wherein the feedback data can be in an unstructured or structured format. Moreover, the system and method discloses associating the feedback data with one or more documents. Additionally, the system and method can rank the one or more documents based from the feedback data. Furthermore, the system and method can provide the ranked one or more documents as search results to a search request.
This Summary is provided to introduce a selection of concepts in a simplified form that are further described below in the Detailed Description. This Summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used as an aid in determining the scope of the claimed subject matter.
Server 104 may be or can include a workstation running the Microsoft Windows®, MacOS™, Unix, Linux, Xenix, IBM AIX™, Hewlett-Packard UX™, Novell Netware™, Sun Microsystems Solaris™, OS/2™, BeOS™, Mach, Apache, OpenStep™ or other operating system or platform. In an embodiment, client 102 may also be a server.
Client 102 can include a communication interface. The communication interface can be an interface that allows the client to be directly connected to any other client or device or that allows the client 102 to be connected to a client, server, or device over network 106. Network 108 can include, for example, a local area network (LAN), a wide area network (WAN), or the Internet. In an embodiment, the client 102 can be connected to another client, server, or device via a wireless interface.
In step 204, the aggregated feedback data can be associated with one or more documents. In an embodiment, the user can transmit feedback data to the search engine in a structured format. In creating a structured format, the search engine can present an user interface (UI) to the user in which the search engine can define what type of information it would like to have from the user. For example, the search engine can create a UI in which it can request that the user provide feedback data regarding a particular document by rating the document on a scale of 1-5 wherein 5 could be the most favorable rating for the document and 1 could be the least favorable rating. Another example could be presenting a UI to the user in which the user can select from any number of buttons that correspond to predefined textual descriptions of a document. The textual description can describe how pleased or how dissatisfied the user is with the document. For example, the UI may contain such terms as “the best possible,” “good,” “fair,” “poor,” or “the worst possible” that the user can select as feedback data regarding a particular document. In yet another example, the search engine can present an UI to the user and ask the user provide feedback data by reordering documents within a search result to show which documents the user believes are more relevant to their search request than other documents. The invention can use a client-side scripting language that can offer a conventional means to allow the user to reorder the documents. For example, the user could click and drag icons, images, or links that represent documents above or below other icons, images, or links of other documents to create a reordered list of documents that correlate to an ascending or descending order of relevance.
In any type of structured format for receiving feedback data in a pre-defined manner, the search engine can associate the feedback data with the particular document because the search already has knowledge of what type of information it will receive from the user. Additionally, if the feedback data was in response to a document included in a list of search results provided by the search engine, the search engine can associate the search query that corresponds to the search results, the feedback data, and document together with each other.
In another embodiment, the user can transmit feedback data to the search engine in an unstructured format. In creating an unstructured format for the user to transmit feedback data, the search engine can present an UI to the user in which the user can freely enter in any type of information that can reveal the user's feelings regarding a particular document. For example, the search engine can present a UI including a text box in which the user can type in whatever comments he/she has as feedback data concerning a particular document.
In an embodiment, the feedback data can be directly correlated to a search query as the feedback data is provided in response to search results received from a search engine that the search engine provided after receiving the search query. For example, the feedback data can be linked to a searching experience wherein the user can provide feedback data that can be associated with a web page that the search engine provided as a search result to the user's search query. Since the feedback data is linked to a search query that the search engine provided search results for, the search engine can associate the feedback data with the particular document and the search query.
In another embodiment, the feedback data may not be directly correlated with a search query and the search engine may not be able to directly link the feedback data with a document or a search query. In fact, a user may be able to provide feedback data that is not directly related to a search experience. More specifically, a user may be able to provide feedback data by contacting the search engine, a separate server that is connected to the search engine, or by contacting the entity that is managing the search engine. For example, the user can send an email as feedback data to the search engine or separate server that can describe their feelings about a particular search result that it received while conducting a search request, or the feedback data can generally describe the user's feelings about a particular document. In another example, the user can call the entity that is managing the search engine and can verbally provide feedback data that can describe their feelings about a particular search result that it received while conducting a search request, or again the feedback data can generally describe the user's feelings about a particular document. The entity can then convert the speech into text using conventional speech-to-text software. In yet another example, the search engine or separate server can retrieve feedback data that may be found on a blog on the Internet. In each example, the search engine or separate server can parse the text using conventional parsing algorithms to associate the feedback data to a particular document and possibly a search query as well.
In step 206, the search engine can store the aggregated feedback data in its index. Again, the index can be, for example, an inverted index. In an embodiment, the feedback data, one or more documents, and search query that were associated together in step 204 can be stored together in the index. In another embodiment, the feedback data can be stored in a separate index from other content that is indexed by the search engine from the network.
In step 208, the search engine can rank one or more documents stored in its index based from the feedback data that it received. The feedback data can be one feature of a plurality of features that can be used to provide better search results for similar search queries that were associated with the feedback data. For example, the text indexed from documents found on the Internet can be one feature to determine the ranking of search results for particular search queries. Another feature can be the text received from the user feedback data. Yet another feature can be the text found specifically from the title of a document. Moreover, another feature can be the ratio of the size of a document to the number of times a particular word is found within the document. It should be noted that any type of feature can be used to determine how to rank one or more documents. The above-mentioned features are only a subset of the types of features that can be used to determine how to rank certain documents, therefore, the invention should not be limited to only those examples.
When ranking a plurality of documents for a particular search query, the search engine can be configured to emphasize the importance of a word found within the feedback data more than the word being found within the context of any other feature. For example, the search engine can emphasize the feedback data that contains a particular word as being three times more important than a document that contains that same word ten times within itself. In helping itself determine the importance of a word relative to what feature it is found in, the search engine can attach different identifiers with particular words of search queries that can inform the search engine as to where the word came from after it is indexed in the search engine. For example, the search engine may use an identifier to signal that the word came from a title of a document, an identifier to show that the word came from the body of the document, or another identifier to show that it came from feedback data.
In other embodiments, search results, including documents, can be ranked based on a comparison of the context of a word within a particular feature. For example, the a word that is found toward the top of a particular item of feedback data can have more importance than the number of times the word in found within the same element of feedback data. The same example can be applied to a document, wherein the number of times a word is found within the document can have more importance than where the word is found within the document. The way in which the search engine is configured to emphasize certain features can determine how the search engine will subsequently rank and provide search results for similar search queries in view of the received feedback data. Once the one or more documents have been ranked, in step 210, the search engine can provide the ranked documents for similar search queries that correspond to the search queries that have already been associated with particular feedback data.
The feedback data that is stored in the search engine's index may have a life span or a staleness rate. For example, in an embodiment, the invention can be configured to emphasize older feedback data less than feedback data that was received at a time closer to when a search request was received from a user. In another embodiment, the staleness rate of the feedback data can vary according to the crawl rate of the document that the feedback data corresponds to. For example, if a document, a web page for example, has a high crawl rate, meaning the data found on the web page is indexed by the search engine frequently (every hour for example), then feedback data that is associated with that web page can become stale and unimportant at the same rate as the web page is indexed. The staleness of the feedback data can then become another feature for determining how to rank one or more documents in view of the feedback data. However, in another embodiment, a document that has a high crawl rate can be deemed to be more important than other documents with lower crawl rates. In such an embodiment, any type of feedback data received that is associated with a document with a high crawl rate can also be deemed important and can become another feature for determining how to rank one or more documents in view of the feedback data.
In an embodiment, the search engine can receive the feedback data in step 306. In another embodiment, a client-side application can be provided to receive the feedback data in step 306. In both embodiments, the search engine and the client-side application can instantly create an item of data based on the feedback data that was received. An item of data can be a second list of search results that includes one or more different documents or a reordered version of the first list of documents. Additionally, an item of data can be a notification that can acknowledge that the search engine or client-side application received the feedback data. When the item of data is a second list of search results, the client-side application or search engine can instantly display the second list when it receives the feedback data from the user at step 308. The client-side application can include a script language that can provide a conventional means to allow the user to reorder or re-rank the documents included in the first list of search results. For example, the user could click and drag icons, images, or links that represent documents above or below other icons, images, or links of other documents to create a reordered list of documents that correlate to an ascending or descending order of relevance. Once the user reorders or re-ranks the documents, the client-side application can instantly display the second list of search results. Alternatively, the search engine can be used to instantly display a second list of search results. The search engine can receive the feedback data, complete steps 204-208 as discussed in
In an embodiment, a client-side application can allow a user to store preferences on the client that can be used along with feedback data to allow the search engine to provide search results that correspond to the user's preferences and feedback data. For example, the user can customize his/her machine to inform a search engine that he/she prefers to have news content, sports content, video content, or image content ranked higher than any other type of content. Additionally, the client side application can inform the search engine of particular physical properties of the user's machine that can be used along with feedback data to provide relevant search results. For example, the client-side application can inform the search engine that the user's machine is a mobile device. The search engine may have certain documents indexed that are better formatted for mobile devices that the search engine can send to the user as search results in view of the user's feedback data.
While particular embodiments of the invention have been illustrated and described in detail herein, it should be understood that various changes and modifications might be made to the invention without departing from the scope and intent of the invention. The embodiments described herein are intended in all respects to be illustrative rather than restrictive. Alternate embodiments will become apparent to those skilled in the art to which the present invention pertains without departing from its scope.
From the foregoing it will be seen that this invention is one well adapted to attain all the ends and objects set forth above, together with other advantages, which are obvious and inherent to the system and method. It will be understood that certain features and sub-combinations are of utility and may be employed without reference to other features and sub-combinations. This is contemplated and within the scope of the appended claims.
