Variable personalization of search results in a search engine
30 claims: 18 independent, 12 dependent
- 1検索クエリの結果をユーザに合わせてパーソナライズする ために、通信ネットワークを介してクライアント・コンピュータと通信するサーバー・コンピュータにおいて実行される 方法であって、 前記クライアント・コンピュータから送られたユーザの関心プロファイルを、前記サーバー・コンピュータで受信するステップと、前記関心プロファイルは該ユーザにとって興味のある少なくとも1つのトピックを表すデータで構成されており、 前記クライアント・コンピュータから送られた複数の文書を検索するための検索クエリを、前記サーバー・コンピュータで受信するステップと、 前記サーバー・コンピュータで受信した前記検索クエリ及び関心プロファイルに基づき、前記サーバー・コンピュータに結合された検索エンジンを介して該受信した検索クエリ及び関心プロファイルに応じた複数の文書を検索するステップと、 前記検索した複数の文書に基づき前記関心プロファイルの複数の度合いに対応する複数組の複数の文書を前記サーバー・コンピュータで生成するステップと、 前記生成された前記複数組の複数の文書のデータを、前記サーバー・コンピュータから前記クライアント・コンピュータに送信するステップであって、これによって、該クライアント・コンピュータの側で、前記複数の度合いのうちいずれか1つの度合いを選択することに応じて前記複数組の複数の文書のうちの1組をユーザに提示することを容易にする前記ステップと を具備する 方法。
- 2前記複数組の複数の文書を前記サーバー・コンピュータで生成する前記ステップは、前記検索するステップにおける検索結果として前記検索エンジンによって各文書に与えられる情報検索スコアに従って各文書の順位付けを決定することを含み、その際に、前記サーバー・コンピュータで生成する前記複数組の複数の文書の各組毎に、当該組内の複数の文書を、各組毎の前記度合いに応じて前記関心プロファイルが各組の順位付けに与える影響が異なるように、前記順位付けを決定することを特徴とする請求項1に記載の方法。
- 3前 記順 位付けを決定することは、 前記 1組の複数の文書中の 少なくとも1の文書の情報検索スコアを、前記ユーザの関心プロファイルに含まれる或る関心事項に対応づけられた前記文書に応じて、該文書に対応づけられたブースト値でスケールすることからなる請求項 2 に記載の方法。
- 4各文書は情報検索スコアを有し、前 記順 位付けを決定することは、 前記ユーザの関心プロファイルに含まれるトピックに対応づけられた各文書に対してブースト値を決定し、該ブースト値及び影響の度合いの分だけ当該各文書の情報検索スコアをスケールすることと、 前記各文書をそれぞれの前記情報検索スコアに応じて順位付けすることと、を具備する請求項 2 に記載の方法。
- 5前 記順 位付けを決定することは、 前記1組の複数の 文書における少なくとも1の文書につき、 当該文書を含むウェブサイトが前記ユーザの関心プロファイルに含まれるトピックに対応づけられているかどうかを判定することと、 当該文書を含むウェブサイトに対応づけられたブースト値で当該文書の情報検索スコアをスケールすることと、を 含む 請求項 2 に記載の方法。
- 6前 記順 位付けを決定することは、前記 1組の複数の 文書における少なくとも1の文書につき、 当該文書が、前記ユーザの関心プロファイルに含まれるトピックに対応づけられた第2のウェブサイトにリンクされた第1のウェブサイトに含まれているかどうかを判定することと、 前記第1のウェブサイトに対応づけられた重み値で当該文書の情報検索スコアをスケールすることと、を 含む 請求項 2 に記載の方法。
- 7前記重み値は前記第2のウェブサイトから前記第1のウェブサイトへのリンク数に基づくことを特徴とする請求項 6 に記載の方法。
- 8前記ユーザの関心プロファイルを前記サーバー・コンピュータで受信する前記ステップは、 前記ユーザが閲覧しているウェブページに対応付けられたクッキーから該ユーザの関心プロファイルを取り出すことと、 前記クッキーから取り出した前記ユーザの関心プロファイルを前記サーバー・コンピュータで受信すること を含むことを特徴とする請求項1乃至7のいずれかに記載の方法。
- 9前記ユーザの関心プロファイルを前記サーバー・コンピュータで受信する前記ステップは、更に、 前記クライアント・コンピュータから受信した前記ユーザの関心プロファイルを、前記検索エンジンに通信可能に結合されたデータベースに記憶することと、 前記データベースから前記ユーザの関心プロファイルを取り出すこと を具備することを特徴とする請求項1乃至7のいずれかに記載の方法。
- 10前記ユーザの関心プロファイルは一組の文字表現から成り、各文字表現は 前記ユーザにとって興味のある複数のトピックの各々 に対応していることを特徴とする請求項1 乃至9のいずれか に記載の方法。
- 11前記ユーザの関心プロファイルは一組のトピック・インデックス番号から成り、各トピック・インデックス番号は 前記ユーザにとって興味のある複数のトピックの各々 に対応していることを特徴とする請求項1 乃至9のいずれか に記載の方法。
- 12前記サーバー・コンピュータから前記クライアントコンピュータに送信される前記生成された前記複数組の複数の文書のデータは、前記生成された前記複数組の複数の文書に含まれる全文書のマッピングと、前記複数組の複数の文書の各組毎に、各組内の各文書の順位付けを示す順位ベクトルとを含むことを特徴とする請求項2に記載の方法。
- 13前記クライアント・コンピュータにおいて前記複数の度合いのうちいずれか1つの度合いを選択するためのユーザ操作を受け付けるステップと、 前記サーバー・コンピュータから送信される前記複数組の複数の文書のうち前記選択された度合いに対応する1組の複数の文書を、前記クライアント・コンピュータに結合されたディスプレイ装置を介してユーザに提示するステップと を具備する請求項1乃至12のいずれかに記載の方法。
- 14前記 ディスプレイ装置を介してユーザに提示する前記ステップは、前記1つの度合いに対応する1組の複数の文書を順位付けに従って表示し、かつ、 前記ユーザの関心プロファイルに含まれる 前記トピック に対応づけられた各文書の隣に、個人向け化されたことを示すアイコンを表示することを具備する請求項 13 に記載の方法。
- 15前記 ユーザ 操作 は、制御操作手段のグラフィック表示物における位置 を 変化 させる操作 であることを特徴とする請求項 13又は14 に記載の方法。
- 16前記制御操作手段はスライダであることを特徴とする請求項 15 に記載の方法。
- 17前記制御操作手段はダイヤルであることを特徴とする請求項 15 に記載の方法。
- 18前記制御操作手段はラジオボタンであることを特徴とする請求項 15 に記載の方法。
- 19前 記順 位付けを決定することは、前記ユーザの関心プロファイルに含まれる 前記 トピックに対応づけられた 各 文書につき、 前記文書を含むウェブサイトを決定することと、 前記ウェブサイトに対応づけられた各トピック毎に、ウェブサイトのブースト値を決定することと、 前記ブースト値の関数として、当該文書の最終ブースト値を決定することと、 前記最終ブースト値で当該文書の情報検索スコアをスケールすることと、を 含む ことを特徴とする請求項 2 に記載の方法。
- 20前 記順 位付けを決定することは、 トピック・ディレクトリが複数のトピックから成り、各トピックが少なくとも1つのウェブサイトに対応づけられており、各ウェブサイトがブースト値を有しており、前記ユーザの関心プロファイルは前記トピック・ディレクトリからの少なくとも1のトピックを含んでおり、少なくとも1の文書につき、前記トピック・ディレクトリからブースト値を決定することと、 前記少なくとも1の文書の情報検索スコアをそのブースト値でスケールすることと、を具備する請求項 2 に記載の方法。
- 21前記クライアント・コンピュータからの前記関心プロファイルの変更を前記サーバー・コンピュータで受け付けるステップを更に具備し、前記変更は、a)前記関心プロファイルに少なくとも1つのトピックを追加すること及びb)前記関心プロファイルから少なくとも1つのトピックを削除することの少なくとも1つと、該変更された関心プロファイルに対応づけられた文書の情報検索スコアを該変更された関心プロファイルに応じて再調整することとを含む請求項1に記載の方法。
- 22検索クエリの結果をユーザに合わせてパーソナライズするためのコンピュータプログラムであって、通信ネットワークを介してクライアント・コンピュータと通信するサーバー・コンピュータに、 前記クライアント・コンピュータから送られたユーザの関心プロファイルを、前記サーバー・コンピュータで受信する手順と、ここで、前記関心プロファイルは該ユーザにとって興味のある少なくとも1つのトピックを表すデータで構成されており、 前記クライアント・コンピュータから送られた複数の文書を検索するための検索クエリを、前記サーバー・コンピュータで受信する手順と、 前記サーバー・コンピュータで受信した前記検索クエリ及び関心プロファイルに基づき、前記サーバー・コンピュータに結合された検索エンジンによって該受信した検索クエリ及び関心プロファイルに応じた複数の文書を検索する手順と、 該検索した複数の文書に基づき前記関心プロファイルの複数の度合いに対応する複数組の複数の文書を前記サーバー・コンピュータで生成する手順と、 前記生成された前記複数組の複数の文書のデータを、前記サーバー・コンピュータから前記クライアントコンピュータに送信する手順であって、これによって、該クライアント・コンピュータの側で、前記複数の度合いのうちいずれか1つの度合いを選択することに応じて前記複数組の複数の文書のうちの1組をユーザに提示することを容易にする前記手順と を実行させるためのコンピュータプログラム。
- 23前記複数組の複数の文書を前記サーバー・コンピュータで生成する前記手順は、前記検索する手順における検索の結果として前記検索エンジンによって各文書に与えられる情報検索スコアに従って各文書の順位付けを決定することを含み、その際に、前記サーバー・コンピュータで生成する前記複数組の複数の文書の各組毎に、当該組内の複数の文書を、各組毎の前記度合いに応じて前記関心プロファイルが各組の順位付けに与える影響が異なるように、前記順位付けを決定することを特徴とする請求項22に記載のコンピュータプログラム。
- 24更に、前記クライアント・コンピュータに、前記複数の度合いのうちいずれか1つの度合いを選択するためのユーザ操作を受け付ける手順を実行させ、かつ、 更に、前記クライアント・コンピュータに、前記サーバー・コンピュータから送信される前記複数組の複数の文書のうち前記選択された度合いに対応する1組の複数の文書を、前記クライアント・コンピュータに結合されたディスプレイ装置を介してユーザに提示する手順を実行させる ことを特徴とする請求項22又は23に記載のコンピュータプログラム。
- 25前記ディスプレイ装置を介してユーザに提示する前記手順は、前記1つの度合いに対応する1組の複数の文書を順位付けに従って表示し、かつ、前記ユーザの関心プロファイルに含まれる前記トピックに対応づけられた各文書の隣に、個人向け化されたことを示すアイコンを表示することを特徴とする請求項24に記載のコンピュータプログラム。
- 26前記ユーザ操作を受け付ける手順及び前記ユーザに提示する手順を前記クライアント・コンピュータに実行させるためのプログラムモジュールが、前記サーバー・コンピュータから前記クライアント・コンピュータに送られるウェブ・ブラウザ・プログラムに含まれることを特徴とする請求項24又は25のコンピュータプログラム。
- 27検索クエリの結果をユーザに合わせてパーソナライズするために、通信ネットワークを介してクライアント・コンピュータと通信するサーバー・コンピュータを含むシステムであって、 前記サーバー・コンピュータが、 前記クライアント・コンピュータから送られたユーザの関心プロファイルを、前記サーバー・コンピュータで受信するための手段と、前記関心プロファイルは該ユーザにとって興味のある少なくとも1つのトピックを表すデータで構成されており、 前記クライアント・コンピュータから送られた複数の文書を検索するための検索クエリを、前記サーバー・コンピュータで受信するための手段と、 前記サーバー・コンピュータで受信した前記検索クエリ及び関心プロファイルに基づき、前記サーバー・コンピュータに結合された検索エンジンによって該受信した検索クエリ及び関心プロファイルに応じた複数の文書を検索する手段と、 該検索した複数の文書に基づき前記関心プロファイルの複数の度合いに対応する複数組の複数の文書を前記サーバー・コンピュータで生成するための手段と、 前記生成された前記複数組の複数の文書のデータを、前記サーバー・コンピュータから前記クライアントコンピュータに送信する手段であって、これによって、該クライアント・コンピュータの側で、前記複数の度合いのうちいずれか1つの度合いを選択することに応じて前記複数組の複数の文書のうちの1組をユーザに提示することを容易にする前記手段と を具備することを特徴とするシステム。
- 28前記複数組の複数の文書を前記サーバー・コンピュータで生成する前記手段は、前記検索する手段における検索の結果として前記検索エンジンによって各文書に与えられる情報検索スコアに従って各文書の順位付けを決定することを含み、その際に、前記サーバー・コンピュータで生成する前記複数組の複数の文書の各組毎に、当該組内の複数の文書を、各組毎の前記度合いに応じて前記関心プロファイルが各組の順位付けに与える影響が異なるように、前記順位付けを決定することを特徴とする請求項27に記載のシステム。
- 29更に、 前記複数の度合いのうちいずれか1つの度合いを選択するためのユーザ操作を受け付ける手段と、 前記サーバー・コンピュータから送信される前記複数組の複数の文書のうち前記選択された度合いに対応する1組の複数の文書を、前記クライアント・コンピュータに結合されたディスプレイ装置を介してユーザに提示する手段と を具備するクライアント・コンピュータ を具備することを特徴とする請求項27又は28に記載のシステム。
- 30前記ディスプレイ装置を介してユーザに提示する前記手段は、前記1つの度合いに対応する1組の複数の文書を順位付けに従って表示し、かつ、前記ユーザの関心プロファイルに含まれる前記トピックに対応づけられた各文書の隣に、個人向け化されたことを示すアイコンを表示することを特徴とする請求項29に記載のシステム。
Independent claims30
58 paragraphs, as filed
<u style="single">Cross-reference of related applications</u>This application is based on 35 USC 35, 119 (e) with respect to Application No. 60 / 557,783 dated March 29, 2004 and Application No. 11 / 002,474 dated December 1, 2004, which are included in the disclosure by reference. Claim priority based on the section. The application also discloses U.S. Patent Application Nos. 10 / 646,331 (filed August 22, 2003) and 10 / 641,489 (filed August 14, 2003) by citation. All of the above applications are by the same person as this application.
The present invention relates to online search engines, and in particular to personalization of search results.
Current online search engine technology is very advanced in its ability to search for documents (web pages or parts thereof, images, etc.) according to the query word. Today's search engines can quickly search for a specific document that matches a query term. However, current search engines often return documents that match a particular query term perfectly but do not reflect the user's potential interests at all. Therefore, if there are two separate users, a user with a high interest in sports and a user with a high interest in politics, the former is about which team performed the drug test, and the latter is about the drug test bill in sports. You'll get exactly the same results for a query, such as a "baseball drug test," even though you want to know. Therefore, it is necessary to provide a mechanism and method for personalizing search results (for each individual) according to the interests of users.
In addition, some users may have specific interests that can be used to process search queries, but others may not like such interests to affect search results. Absent. Therefore, it would be desirable to provide a mechanism and method that allows users to variably adjust how much they want to influence their interests in the results of a search query.
The present invention allows a user to variably adjust how much his or her personal interests affect the results of a particular search query. The present invention has a plurality of examples including a system that provides a search engine, a method for processing a search query, and a user interface that allows the degree of influence to be variably adjusted.
The present invention provides a mechanism for personalizing (personalizing) search results so that users can see them. The user defines a set of interests, such as a set of topics or categories of interest to the user, as an interest profile. The user then submits the search query to the search engine. The search engine searches for documents related to the query and presents them as search results. Initially (but not necessarily) the search results are ranked regardless of the user's interests. As a result, the user can see the relevance of the document unaffected by his or her personal interests. In a preferred embodiment, this is done by directly manipulating a graphic control element, i.e. a personalized control (control operating means for personalization). This graphic element can be represented by various graphic representations and is displayed as a slider in one embodiment. By adjusting this control element, the degree of influence of the user's interest as represented in the user's interest profile on the ranking of results is adjusted. When the control element is repositioned, the search results are re-ranked, which allows the user to recognize how much their interest profile affects the ranking of the results.
In one embodiment, the ranking of search results serves to increase the relevance score of the documents associated with the topic directory. Each topic in a directory is associated with a set of documents (such as a website) that appear to be related to that topic. Each document (again, the website) is given a boost value that indicates how relevant the site is to the topic. Every page in a site is given a boost value for that site. The boost value can be interpreted as a multiplier that represents the degree of relevance of the site to the topic. During query processing, the above set of documents is returned and ranked regardless of their inclusion in the topic directory. Further, the individual score for each document is determined by adjusting the basic relevance score of the document, if any, by multiplying the relevance score by the corresponding boost value, thereby and the query of the document. Scale (adjust) relevance. In general, the relevance score is only increased for documents that belong to the sites listed in the directory, not for other documents in the search results. Then, the re-ranking of documents according to the increased relevance score is from the position where it does not affect (the position where the boost value does not affect the relevance score) to the position where the effect is maximized, that is, the position of each document. It is a function of the position where the boost value is 100% applied to the relevance score, and the position of the personalized control, which is one of the intermediate application levels. Therefore, the user can quickly reorder the documents in the search results by adjusting the position of the controls.
Examples of the system are topic directories that provide topics (categories) to users, search engines and indexes that search websites indexed according to user queries, and search results according to the user's interests. Includes personalized controls for determining the ranking of at least one (preferably a set) of. An example user interface is a topic directory with various topics and graphic controls (such as checkboxes) that allow the user to select topics of interest, and the user adjusts the ranking of search results according to their interest profile. Includes graphic controls that allow you to.
Hereinafter, the present invention will be described with reference to various drawings and technical information. The drawings show various embodiments of the invention for illustration purposes only. Those skilled in the art will readily appreciate from the following description that other embodiments of the structures, methods and functions illustrated and described can be used without departing from the principles of the invention.
Figure 1 shows the search engine entrance (landing page) 100. The search engine includes a query entry field 102 for the user to enter a search query term. The search button 106 submits to a search engine to process the query. When a user accesses a search engine for the first time, a user profile for personalizing the search results has not yet been defined, and a message to that effect is displayed. Users can access profile edit link 104 and set their own personal interests.
Click the profile edit link 104 to access page 200 shown in Figure 2. This page shows top-level topics 204 in topic directory 202. Topics in this directory can be, for example, topics defined in the Open Directory project at http://dmoz.org/ or other topic directories. (In this disclosure, the terms "topic" and "category" are interchangeable terms, as are "subtopics" and "subcategories"). Links to top-level topics can have a set of subtopics, as indicated by arrow 206. The user can start the search at any time by clicking the search start button 208. Clicking on a particular top-level topic takes the user to a page related to that topic's subtopics.
Figure 3 shows the result of the step in which the user clicked on the topic "Computer" on page 200. As shown in Figure 3, subtopic 302 of the topic "Computer" is displayed in the right column. Next to each subtopic 302 is a checkbox 314. The user can indicate interest in a particular subtopic by selecting check box 314. Alternatively, the user can select the entire subtopic, ie the "Computer (General)" checkbox 316. This indicates that the user is interested in all the subtopics of the topic "Computer". A set of topics selected by the user can be referred to as a user profile, personalized profile, or interest profile. On the left side of Figure 3, after the user has selected one or more topics of interest, these topics are listed in interest column 306. Next to each topic name is a delete link 308 that allows the user to delete each topic. The user can also delete all topics by selecting the "Delete all preferences" link 310.
The subtopic itself can have any number of additional level subtopics. For example, FIG. 4 shows that the subtopic "game" further has subtopics 402 such as "board game" and "card game". When the user clicks the subtopic arrow 206 next to the label of the subtopic "Game", the above subtopic is shown to the user. The user can also select the entire subcategory by selecting check box 404, or select check box 406 for individual subtopics. In one embodiment, in addition to the top-level topic 204, there are two sub-topic levels. This allows the topic to be covered broadly and finely enough without unnecessarily complicating calculations.
Looking back at Figure 3, users show an interest in computers in general. Clicking the search start button 208 will bring up the search page 100 again, which no longer displays a message prompting the user to create a personalized profile. At this point, the user enters a search query, here "Stanford", in the search input field, as shown in Figure 5. The search engine searches the index and provides the search results. Figure 6 shows the first search result page 600. Here, the search result 602 consists of a set of documents 604 indicated by URLs, a link 615 to the page, and a page text excerpt 604 to show the user a part of the context. Near the top of the page is a graphic display of personalized controls 608. In the context of this application, "document" is any type of medium that can be indexed and searched by search engines, including web pages, images, multimedia files, text documents, PDFs and the like. The use of the generic term "document" does not limit the invention.
The personalized control 608 allows the user to adjust the ranking of each document / page in the search results based on their own interests, preferably as represented by a personalized profile. The personalization control can be described as adjusting the degree of influence or contribution that the user's interest profile has on the ranking of search results. FIG. 6 shows a control 608 in an embodiment in which the control 608 is a slider type including a control handle 610 that is moved along a straight line between a minimum value 612 and a maximum value 614. This type of control 608 is generally very sensory and the user can easily understand how it works. In another embodiment, there is also a dial-type control that can be variably adjusted between the minimum and maximum values, and radio buttons that correspond to a predetermined adjustment value (for example, 0%, 20%, and 40%, respectively). , 80%, 100% corresponding to 5 radio buttons). Those skilled in the field of user interface design will find that other types of controls can be used to achieve essentially the same effect.
In the example shown in FIG. 6, the personalized control 608 is set to the minimum value 612. Therefore, the search results are related to the query and, by this reference, the whole and all of the teachings are included in the disclosure. PageRank as disclosed in US Pat. No. 6,285,999, "Methods for ranking nodes in linked databases." It is ranked based on query independent information such as. The search results are mostly related to Stanford University.
FIG. 7 shows the ranking of search results 602 after the user adjusts the slider 610 to an intermediate position approximately halfway between the minimum and maximum values. Here, each document 604 is re-ranked, with those that are more relevant to the user's interests (identified above as "computer (general)"). These documents are indicated by a personalized icon 618 next to the link text 615 so that the user can identify which documents are more relevant to their interests. If the slider 610 is in the middle position, every other document will be the personalization result, but this is not a requirement. In addition, the text in Excerpt 619 contains text related to computers (such as "hacking"), but the term is not included in the original search query.
Next, FIG. 8 shows the ranking of the search result 602 after the user adjusts the slider 610 to the maximum value. First, in this example, all of the documents 604 in search result 602 reflect the user's personalized profile and are labeled with the personalized icon 618. Second, the documents in the search results are related to computers and Stanford University, so while related to the query "Stanford", they reflect the user's interest in computers. For example, the second-to-top results relate to "hacking" in Stanford, and the following three documents specifically relate to Stanford's various computer science projects. So the third document is about Stanford's Institute of Artificial Intelligence, the fourth is about Stanford's Center for Linguistic Informatics, and the fifth is about Stanford's databases. Similarly, the rest of the documents relate to the "computer" that the user is interested in. Also, to the Stanford University home page, which was the first (most relevant) search result when personalization was not applied (Figure 6) and when moderate personalization was applied (Figure 7). The link does not appear in the top eight results.
In the examples shown in FIGS. 6 to 8, the user can dynamically adjust the ranking of documents in real time without the search engine recalculating the search results. This is even more effective for the user to search for the search result.
As shown in FIGS. 6-8, the profile edit link 616 allows the user to edit the interests of the user profile at any time. Figure 9 shows the user returning to topic directory page 200, as shown in Figure 2, removing the interest in the topic of the current computer and creating a new interest in "music". Then, the user selects the search start button 208 again. The search engine provides a set of updated search results, but the search results document is the same as before because the query terms haven't changed. However, at the time of the latest search, the user has positioned Control 610 for full personalization, so search engines have applied this setting to match the current (and new) personalized profile, this time "Music". To rank the documents. Figure 10 shows the dramatic difference this results in search results. Here, the top six results are all related to music and art, and none of the previous computer-related results are at the top of the search results list. This indicates that the user's personalized profile can significantly change the ranking of a set of documents.
In one embodiment of the present invention, the search results are personalized by the following method. Those skilled in the art of search engine design will be able to easily understand the basic structure of a search engine and the processing in which this method is implemented.
1) When the user selects a topic (category), the topic is added to the cookie containing the user's interest list IN (IN stands for "interest"). The topic of interest (topic, heading, item) is an index inserted as a text token (eg, a textual representation such as "computer") or in a directory, with each topic / subtopic having an index number. Can be expressed as. When a user visits the site, the system searches this interest list from cookies. Alternatively, the site may display a user registration page where the user registers with a user name and password, and the user's interests are stored in a search engine site in the database and searched in response to the user's login.
2) Each time the user issues a search query Q, the list IN is sent to the front-end web server.
3) Next, when the front-end web server issues two requests, (Q null) and (Q IN) to the back-end search engine, the search engine sends two sets of results to the front-end web server. Returns R_base (normal) and R_in (personalized).
4) Since a basic query (Q null) is the same as searching for Q on a search engine without personalization, R_base is the search result that the search engine would normally return to the query. As mentioned above, the search results are relevant to the query and other factors, such as US Pat. No. 6,285,999, "How to rank nodes in a linked database," which is included in the disclosure in its entirety and in its entirety by this reference. It is ranked according to PageRank etc. as disclosed in.
The personalized query (Q IN) contains the user's list of interests. Search engines follow a process control flow similar to a normal query, but activate a personalization element while the scoring feature is running, which "boosts" documents related to a set of interest INs.
In one embodiment of the invention, the boost is performed by a set of maps of the format shown below (each map corresponds to each topic / subtopic in the directory).
{Site <u style="single">――></u> boost}
So, for example, the category "health" has a map of the following format.
{Site <u style="single">――></u> boost}
For example, it contains the following entry:
nih.gov <u style="single">――></u> 5.8
cdc.gov <u style="single">――></u> 7.9
med.stanford.edu <u style="single">――></u> 3.5
nih.gov, cdc.gov, and med.Stanford.edu are various sites that have been determined (either manually or automatically) to be relevant to the topic "health". In other words, for each topic in the directory, there is a set of sites that are determined to be related to that topic, and a boost value is defined for each of these sites.
The boost for sites listed in the topic directory is typically determined as follows:
a) The node in the graph is a site (basically a page on the same host),<u style="single">One</u>From one site to another<u style="single">Link to the page</u>Based on the number of pages<u style="single">The edges between the nodes</u>Weighted<u style="single">ing,</u>A "site graph" is generated. This same type of graph is used to calculate all topic boost maps.
b) For each topic in the directory, eg "health", a number of sites are selected as the "start site" S0 whose home page is listed in the open directory. For example, for universities such as Stanford, the start site can be selected as any site ending with ".Stanford.edu.".
c) Perform the calculation in the following two paths:
i) First, define a set of sites S1 that are more closely linked to the sites of S0, each site being weighted according to how closely it is linked by the site of S0.
ii) Next, define a set of sites S2 that are more closely linked to the sites of S1 and each site is weighted in the same way as S1.
d) The S2 site is boosted with the attached weight.
When a personalized query joins a set of interests (eg, "health", "music", etc.), the personalized element examines each map and the personalized element IRs the query results (ie, documents) contained in that map. As a result of boosting search scores, those search results are ranked higher. The IR score is an information retrieval score provided by a search engine for each document and can be a function of document relevance and other factors such as PageRank.
For example, if one of the results of query Q is http://www.med.stanford.edu/research and contains IN "health", then that page is on the med.stanford.edu site. The above result gives a multiplication boost value of 3.5 for that information retrieval score. For multiple interests, multiply the values in the corresponding maps to get the overall value. An example for determining site boost in a directory boost map is described in more detail in US Patent Application No. 10 / 646,331 contained in the disclosure by this reference.
6) The front-end web server receives R_base and R_in. R_base represents the result 602 shown to the user when personalized control 608 is set to 0 (or minimum position 612). R_in represents the result 602 shown to the user when the personalized control 608 is set to its maximum position 614, eg 10. In some embodiments, the control has 11 positions, but any number may be used. The order shown to the user from 0 to 10 is interpolated, which uses R_base and R_in to set the control 608 to the maximum position from the position when the control 608 is set to the minimum position. The personalized element creates a rank (R_1 ~ R_9) by moving each result along the virtual path to the position of the case.
7) How the slider effect is supplied to the user:
The front-end web server adds the results of R_in that are not in R_base to R_base to get a list of all unique results (in R_base or R_in) called R_all. The front-end web server then assigns a continuous identification number as an identifier to the results contained in R_all. The front-end web server then sends the following two to the user's browser:
a) Mapping {id-> result} of each result contained in R_all, including the identification number for each document contained in the result, and
b) N vectors indicating the ranking that the user should see by the identification number of the result, where N indicates the number of slider positions, and each vector indicates the ranking of the result based on the degree of personalization. In this example, there are 11 vectors corresponding to positions 0 to 10 of control 108. In one embodiment, vector 0 reflects the ranking of unpersonalized results if the local ID starts with R_0 [0,1,2,3,4,5,6,7,8,9] , 10]. Vector 10 shows the ranking of the results in a fully personalized R_in. For example, if documents 15 and 17 are documents in the topic directory relevance map where the information retrieval score is increased according to the map boost value and the value corresponding to the position of control 608, [15.17,5,0, 1,2,3,4,8,9,10]. By sending N ranking vectors to the client, the client's browser can quickly recalculate the ranking without imposing a computational burden on the server.
8) Finally, the Javascript code on the resulting page causes the user's browser to see one of the 11 ranking vectors as appropriate as the position of control 608 is changed by the user. Search results can be re-ranked using. One embodiment of steps 6-8 above is described in more detail in US Patent Application No. 10 / 646,331, which is included in the disclosure by this reference.
An example in which the present invention can be realized has been described in detail. Those skilled in the art will appreciate that the present invention can be practiced in other embodiments as well. First, the names of the components, the use of capital letters for terms, attributes, data structures, or other programming or structural aspects are neither essential nor important, and the mechanisms that implement the present invention or its features are: It can have a different name, form, or protocol. Further, as described above, the system can be implemented by a combination of hardware and software, or by hardware elements as a whole. Further, the distribution of functions among the various system components described here is merely an example and is not essential. Functions performed by a single system component may be performed by a plurality of components, or functions performed by a plurality of components may be performed by a single component.
Part of the above description demonstrates the features of the invention by algorithms and symbolic representations of the processing performed on the information. Descriptions and representations by such algorithms are means used by those skilled in the art to most effectively convey the content of the processing to those skilled in the art in the field of data processing technology. Such processing is explained functionally or logically, but is carried out by a computer program. Furthermore, it has been found that it may be convenient to refer to such an array of processes as a module or by a functional name without losing generality.
As is clear from the above description, unless otherwise specified, the description using terms such as "calculate", "determine", and "specify" in the entire detailed description is computer memory, registers, or other information storage. It refers to the operation and processing of a computer system or similar electronic computing device that processes and converts data represented as physical (electronic) quantities in a device, transmission device, or display device.
Aspects of the present invention include processing steps and instructions described here by the algorithm. The processing steps and instructions of the present invention can be embodied by software, firmware, or hardware, and when embodied by software, they are resident on various platforms used by the real-time network operating system. It can be downloaded for operation from the platform.
The present invention also relates to an apparatus for performing the processes disclosed herein. The device may be configured specifically for the intended purpose, or may consist of a general purpose computer that is selectively started or reconfigured by a computer program stored in the computer. Such computer programs include, for example, any type of disk such as a floppy disk, optical disk, CD-ROM, optical magnetic disk, etc., each connected to a computer system bus, read-only memory (ROM), random access. It can be stored on a computer-readable storage medium such as memory (RAM), EPROM, EEPROM, magnetic or optical card, any type of medium suitable for storing electronic instructions. Those skilled in the art of integrated circuit design and video codecs can understand that the present invention can be easily constructed with various integrated circuits such as application specific integrated circuits (ASICs) based on functional and structural explanations. Let's do it. Further, the present invention can be incorporated into various video coding devices.
The algorithms and indications presented here are not specific to any particular computer or other device. Various general-purpose systems can be used with programs that follow the teachings herein, and it may be more convenient to build more specialized equipment to perform the required method steps. The structure required for such various systems is clear from the above description. Furthermore, the present invention does not describe a particular programming language. Various programming languages can be used to carry out the teachings herein described above, and specific languages are referenced to disclose the enablement requirements and best embodiments of the invention.
Finally, the terms used herein are primarily selected for readability and educational purposes, and may not be selected to delineate or limit the subject matter of the invention. is there. Therefore, the disclosure of the present invention is exemplary and does not limit the scope of the invention.
<figref num="1">An input web page for search engines.</figref><figref num="2">A web page for editing a user's personal profile about a topic.</figref><figref num="3">Another web page showing topics and subtopics.</figref><figref num="4">Another web page showing subtopics.</figref><figref num="5">Indicates a search engine input web page containing the query "Stanford".</figref><figref num="6">Shows the first search results for the query "Stanford" and a control 608 that controls the degree of influence of the user's interest profile in ranking the search results.</figref><figref num="7">The ranking of the search results in FIG. 6 after the user adjusts the control 608 to the intermediate position is shown.</figref><figref num="8">The ranking of the search results in FIG. 6 after the user adjusts the control 608 to the maximum position is shown.</figref><figref num="9">Shows the web page of a topic after the topic "Computer" has been removed by the user and the topic "Music" has been selected.</figref><figref num="10">The ranking of the search results in Figure 6 following the change in interest in the topic shown in Figure 9 is shown.</figref>
10 sheets
Sheet 1 Sheet 2 Sheet 3 Sheet 4 Sheet 5 Sheet 6 Sheet 7 Sheet 8 Sheet 9 Sheet 10
Every citation, both ways
| Document | Relation | Office |
|---|---|---|
| US20020078045A1 | Cites | United States of America |
| JP2002531895A | Cites | Japan |
| JP2002529859A | Cites | Japan |
| JP200470504A | Cites | Japan |
| JP200232401A | Cites | Japan |
| 灰原 清太郎,サイト評価情報を用いたWWW検索表示方法,情報処理学会論文誌 第40巻 No.SIG6(TOD3),日本,社団法人情報処理学会,1999年 8月15日,第10巻,第22頁乃至第30頁 | Non-patent | – |
30 members in 6 offices
Members30
| Document | Office | Kind | |
|---|---|---|---|
| US2005216434A1 | United States of America | A1 | |
| WO2005098611A2 | World Intellectual Property Organization (WIPO) | A2 | |
| WO2005098611A2 | World Intellectual Property Organization (WIPO) | A2 | |
| EP1741024A2 | European Patent Office (EPO) | A2 | |
| KR20070007150A | Republic of Korea | A | |
| KR20070007150A | Republic of Korea | A | |
| WO2005098611A3 | World Intellectual Property Organization (WIPO) | A3 | |
| WO2005098611A3 | World Intellectual Property Organization (WIPO) | A3 | |
| JP2007531160A | Japan | A | |
| CN101080709A | China | A | |
| EP1741024A4 | European Patent Office (EPO) | A4 | |
| CN100593168C | China | C | |
| US7716223B2 | United States of America | B2 | |
| US2010169297A1 | United States of America | A1 | |
| CN101814079A | China | A | |
| JP4664355B2This record | Japan | B2 | |
| US8180776B2 | United States of America | B2 | |
| KR20120073324A | Republic of Korea | A | |
| KR20120073324A | Republic of Korea | A | |
| US2012221568A1 | United States of America | A1 | |
| KR20120130016A | Republic of Korea | A | |
| KR20120130016A | Republic of Korea | A | |
| US2013103683A1 | United States of America | A1 | |
| CN101814079B | China | B | |
| KR101291291B1 | Republic of Korea | B1 | |
| KR101291291B1 | Republic of Korea | B1 | |
| KR101296211B1 | Republic of Korea | B1 | |
| KR101296211B1 | Republic of Korea | B1 | |
| US8874567B2 | United States of America | B2 | |
| US9058364B2 | United States of America | B2 |
14 legal events, as the office reported them to INPADOC
Over the term
Point at a mark for the eventEvents
| Event | Code | |
|---|---|---|
| Cancellation because of no payment of annual feesLAPS | LAPS | |
| Receipt of annual feesJAPANESE INTERMEDIATE CODE: R250R250 | R250 | |
| Renewal fee payment (event date is renewal date of database)FPAY | FPAY | |
| Certificate of patent or registration of utility modelJAPANESE INTERMEDIATE CODE: R150R150 | R150 | |
| First payment of annual fees (during grant procedure)JAPANESE INTERMEDIATE CODE: A61A61 | A61 | |
| Written decision to grant a patent or to grant a registration (utility model)JAPANESE INTERMEDIATE CODE: A01A01 | A01 | |
| Written decision to grant a patent or to grant a registration (utility model)JAPANESE INTERMEDIATE CODE: A01A01 | A01 | |
| Decision of grant or rejection writtenTRDD | TRDD | |
| Written amendmentJAPANESE INTERMEDIATE CODE: A523A521 | A521 | |
| Written submission of copy of amendment under section 19 (pct)JAPANESE INTERMEDIATE CODE: A524A524 | A524 | |
| Written permission of extension of timeJAPANESE INTERMEDIATE CODE: A602A602 | A602 | |
| Written request for extension of timeJAPANESE INTERMEDIATE CODE: A601A601 | A601 | |
| Notification of reasons for refusalJAPANESE INTERMEDIATE CODE: A131A131 | A131 | |
| Written request for application examinationJAPANESE INTERMEDIATE CODE: A621A621 | A621 |
Numbers
- Publication
- 4664355
- Application
- 2007506227
Titles2
- Japanese
- 検索エンジンにおいて検索結果を可変的にパーソナライズ化すること
- English
- Variable personalization of search results in search engines
Classification
- CPC, 4
- G06F16/9535
- G06F16/24578
- G06F17/00
- Y10S707/99936
- IPC, 2
- G06F17 30
- G06F7 00
