Patent Application
20130254189
The core of the World Wide Web (WWW) comprises several billion interlinked web pages. Accessing information on almost any of these web pages would be essentially impossible without the aid of systems that enable a user to search for specific text, or textual identifiers. Indeed, such systems, generally known as “search engines,” have increased in popularity as the WWW has grown in size.
Traditionally, a search engine comprises an initial page providing the user with a mechanism for entering one or more words, characters, or phrases, known as the “search string” or the user's “query.” The search string represents the words, characters or phrases that the user wishes to find in one or more of the web pages that comprise the WWW. The search engine will then reference a database comprising the content of a myriad of web pages, seeking to identify one or more web pages that contain the search string that the user entered. More advanced search engines can also apply common linguistic permutations to the words or phrases that the user is searching for in an effort to provide a more complete result. Thus, if the user searched for the word “computers,” the search engine could also search for the singular form “computer,” or even the verb form “computing,” so as not to exclude, for example, a web page referencing a single “computer” but not multiple “computers.”
To generate a database which can be searched for the user's search string, search engines commonly employ automated processes known as “crawlers” to read information from a web page, follow one of the links in the web page to another web page, read information from that web page, and so forth until the crawler reaches a web page that comprises no links, in which case the crawler can return to a prior, linking, web page and proceed with a different link. In such a manner, the crawler traverses web pages of the WWW in an orderly manner, returning the information from the web page back to the search engine for storage. The search engine then stores the information in an optimized format to reduce the amount of storage space used, and to improve searching efficiency.
The search results that are generated by a search engine in response to a user's entry of a search string are generally presented to the user in the form of a series of links to the web page determined relevant to the user's query. Clicking on any of the links will enable the user to visit the web page identified by the search engine. However, to avoid requiring the user to click on each result, a search engine can also provide a short excerpt from the web page, illustrating to the user how the search string is used in the web page. A user can then obtain some context information without clicking on each result.
Many web pages do not contain any text, or contain very little text, that can be read by a search engine crawler. For example, a web page can embed a programmatic element, commonly referred to as a “control,” to dynamically display content that, while presenting words on the screen for the user, does not comprise data in a textual format. In such a case, traditional search engines would not display any excerpt, as none could be obtained from the web page by the crawler.
A web page without content in a textual format could, nevertheless, be relevant to a user's query. One mechanism for determining the relevance of a particular web page to various search terms is to track the terms used by other web pages to refer to that web page. Consequently, the original web page need not have any content in a textual format in order to be considered relevant to a user's query. Indeed, the search engine may provide a web page as a result, not because that web page itself contains any of the terms of the search string, but rather because one or more other web pages reference the result web page using terms from the search string. In such a case, no excerpt could be displayed to the user because, while the web page selected as a search result may or may not comprise some text, that text does not include the terms of the user's query.
As used herein the term “page” is meant to include any presentation of information that is intended to be considered as a whole. Thus, while a “web page” is one common type of “page” the term “page” is not intended to be limited to documents in the Hyper-Text Markup Language (HTML). Instead, the term “page” is meant to include both HTML documents and documents and subsections of documents in other formats, including, but not limited to, word processing document formats, spreadsheet document formats, animation, audio or video document formats, presentation software formats and universal (or “portable”) document formats. The elements used by one page to reference another will be termed herein as “anchor text,” though, as described below, the use of the term “text” is not meant to limit the description to text-based formats. In a linked environment, such as the WWW, a link to another page can be represented on the linking page as one or more textual characters, generally identified via font, color, underlining, bolding or some combination thereof, or other non-textual elements, including images, graphics or other data. The visual representation of the link from one page to another is the “anchor text”, and can often provide the reader of the linking page with some information about the linked-to page. For example, the anchor text “computer” in one page would likely link to another page that had at least some connection to the term “computer.” Thus, search engines can deem a page relevant to a user's query simply because one or more other pages reference that page by using anchor text that is similar, or identical, to the user's query. The page deemed relevant to the user's query under such a system, however, may not itself contain the user's query.
To avoid user confusion, and to provide context for the results generated by the search engine, additional information can be provided to the user in conjunction with the links to the pages deemed relevant by the search engine. More particularly, a page that is determined to be relevant to the user's search term because of the anchor text in other pages that links to that page, can be presented by including that anchor text. In one embodiment, if the page that comes up as a search result has no information that can be excerpted, then the anchor text linking to that page can be displayed. In an alternative embodiment, if the page that comes up as a search result does have information that can be excerpted, or if it does have describing information that can be presented, then such information can be displayed in addition to the anchor text from the referring page or pages that contains elements of the user's query. In yet another alternative embodiment, the anchor text can itself include excerpts to provide additional context.
To more efficiently determine which referencing text and page to display, together with the search result page that is being referenced, the referencing information can be compiled and stored. A crawler mechanism can provide both the information from the crawled page, and the anchor text to the crawled page from a prior page. Upon amalgamating the information from multiple crawler mechanisms, multiple anchor texts to a single page may be identified. Such anchor texts can be sorted based on their specific wording. In many cases, multiple pages may refer to a referenced page using identical anchor tag wording. For each wording, a determination can be made of the “static rank” of the pages using that exact wording to refer to the referenced page, where the “static rank” of the page is the score, rank or other numerical value that is assigned by a search engine to a page independently of any specific user query. The average static rank of all such pages, and the maximum static rank of such pages can be used to identify common anchor text wording for a particular referenced page. In one embodiment, several most common anchor text wordings, together with additional information, can be stored with the referenced page.
Once a user enters a particular search string, the result pages can be identified by the search engine. For each result page that is to be displayed to the user, a determination can be made whether a suitable excerpt from the page can be provided. If a suitable excerpt cannot be provided, some descriptive information or excerpt can be provided and, in addition, referencing information, including the anchor text, can be provided illustrating the use of some or all of the search string on pages referring to the result page.
The specific anchor text wording provided can be selected based on one or more factors. More particularly, the anchor text wording can be selected from among the several most common anchor text wordings that would have been stored with the result page. Factors that can be considered include the number of terms which the anchor text shares with the search string and the overall similarity of the anchor text to the search string. Additional factors include the language of the anchor text as compared to the search string and the results page, and the length of the anchor text. Still further factors can be based on the static rank of the pages that contain the anchor text. Once the anchor text wording is selected, the page to which the user will be connected if they select the anchor text from the search results page can be selected based on similar factors, or it can simply be the page with the highest static rank that uses the anchor text.
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 to limit the scope of the claimed subject matter.
Additional features and advantages will be made apparent from the following detailed description that proceeds with reference to the accompanying drawings.
The following detailed description may be best understood when taken in conjunction with the accompanying drawings, of which:
The following description relates to providing additional context for search results. In one embodiment, pages that either have no text, or whose text does not contain the search terms, may have been selected and presented as search results because they are referred to by other pages in a manner that uses one or more of the search terms. In such cases, the user can be presented with the referencing information, including the anchor text, in order to relate the user's query to the result. In addition, the anchor text can be presented as a link to one or more referencing pages using that anchor text should the user wish to visit those pages. In an alternative embodiment, such referencing information can be presented even if the search result pages do have some or all of the search terms. In such a case, the referencing information can provide further context.
The techniques described herein focus on the determination of when to display referencing information and what referencing information to display. In one embodiment, the referencing information displayed comprises the anchor text used to link to the page being presented as a search result. In another embodiment, the referencing information comprises the anchor text and additional excerpts from the referencing page. The anchor text displayed can be selected based on one or more factors that include, but are not limited to, the length of the anchor text, the static rank of pages using the anchor text, including the average static rank of those pages and the maximum static rank, the similarity of the anchor text wording to the query wording, including the number of terms they have in common and their respective languages, the differences between the anchor text, the query wording and the results page, and other factors.
The determination of when to display referencing information can, in one embodiment, be based on the ability to obtain an excerpt from a search result page that comprises at least some of the search string. In another embodiment, the referencing information can be displayed even if information about the search result page is displayed instead of an excerpt from the page. To render the display of the referencing information more efficient, specific anchor text wordings, and other referencing information can be stored together with the page that is being referenced. In one embodiment, the specific anchor text wordings are selected for storage based on one or more factors, including, but not limited to, the number of times the specific anchor text wording is used in the referring page, the average static rank of all of the pages using that specific anchor text wording, and the maximum static rank out of all of those pages.
Although not required, the description below will be in the general context of computer-executable instructions, such as program modules, being executed by a computing device. More specifically, the description will reference acts and symbolic representations of operations that are performed by one or more computing devices or peripherals, unless indicated otherwise. As such, it will be understood that such acts and operations, which are at times referred to as being computer-executed, include the manipulation by a processing unit of electrical signals representing data in a structured form. This manipulation transforms the data or maintains it at locations in memory, which reconfigures or otherwise alters the operation of the computing device or peripherals in a manner well understood by those skilled in the art. The data structures where data is maintained are physical locations that have particular properties defined by the format of the data.
Generally, program modules include routines, programs, objects, components, data structures, and the like that perform particular tasks or implement particular abstract data types. Moreover, those skilled in the art will appreciate that the computing devices need not be limited to conventional personal computers, and include other computing configurations, including hand-held devices, multi-processor systems, microprocessor based or programmable consumer electronics, network PCs, minicomputers, mainframe computers, and the like. Similarly, the computing devices need not be limited to a stand-alone computing devices, as the mechanisms may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote memory storage devices.
With reference to
Also illustrated in
The user, such as a user of web browser 11 on personal computing device 10, can provide a search string to the search engine 41 through the front end 46, which comprises both the initial page enabling a user to enter search terms, and subsequent pages that are generated with some or all of the results. Upon receiving a search string, the search code 44 can search through the database 50 and identify the pages that are relevant to the search string. In addition, the search code 44 can also rank the identified pages such that those pages deemed most relevant would be the first to be presented to the user. The snippet generator 45 can likewise search the database 50 and identify, for each page deemed relevant by the search code 44, an excerpt from that page illustrating the page's relevance. Such excerpts can be presented to the user via the front end 46 and are commonly referred to as “snippets.”
Turning to
The exemplary computing device 100 can include, but is not limited to, one or more central processing units (CPUs) 120, a system memory 130, and a system bus 121 that couples various system components including the system memory to the processing unit 120. The system bus 121 may be any of several types of bus structures including a memory bus or memory controller, a peripheral bus, and a local bus using any of a variety of bus architectures. By way of example, and not limitation, such architectures include the Peripheral Component Interconnect (PCI) bus and various higher speed versions thereof, the Industry Standard Architecture (ISA) bus and Enhanced ISA (EISA) bus, the Micro Channel Architecture (MCA) bus, and the Video Electronics Standards Associate (VESA) bus. The computing device 100 can optionally include graphics hardware, including, but not limited to, a graphics hardware interface 190 and a display device 191.
The computing device 100 also typically includes computer readable media, which can include any available media that can be accessed by computing device 100 and includes both volatile and nonvolatile media and removable and non-removable media. By way of example, and not limitation, computer readable media may comprise computer storage media and communication media. Computer storage media includes media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can be accessed by the computing device 100. Communication media typically embodies computer readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism and includes any information delivery media. By way of example, and not limitation, communication media includes wired media such as a wired network or direct-wired connection, and wireless media such as acoustic, RF, infrared and other wireless media. Combinations of the any of the above should also be included within the scope of computer readable media.
The system memory 130 includes computer storage media in the form of volatile and/or nonvolatile memory such as read only memory (ROM) 131 and random access memory (RAM) 132. A basic input/output system 133 (BIOS), containing the basic routines that help to transfer information between elements within computing device 100, such as during start-up, is typically stored in ROM 131. RAM 132 typically contains data and/or program modules that are immediately accessible to and/or presently being operated on by processing unit 120. By way of example, and not limitation,
The computing device 100 may also include other removable/non-removable, volatile/nonvolatile computer storage media. By way of example only,
The drives and their associated computer storage media discussed above and illustrated in
Of relevance to the descriptions below, the computing device 100 may operate in a networked environment using logical connections to one or more remote computers. For simplicity of illustration, and in conformance with the exemplary system 99 of
As indicated previously, network communication between a user and the search engine 41 occurs via pages generated by the front end 46 and displayed for the user by web browser 11. Turning to
Some pages, however, have no text at all, or no text that contains the search string. For example, some pages can be mostly comprised of embedded controls that, while they may present user-visible text when the page is viewed by the user, do not comprise information such as could be read and understood by the crawler 42. Nevertheless, such pages may be deemed relevant to the search string because other pages use anchor text similar, or identical, to the search string to refer to those pages. Search result 240 illustrates an example of such a page, containing no text in common with the search string. In such a case, as will be described further below, the snippet generator can attempt to generate a query-independent snippet. However, such a snippet may not convey sufficient information to explain to the user why the page was deemed relevant to the user's search. In the case illustrated in
However, as illustrated in
As indicated above, if the query text is not present in the page deemed relevant to the search result, the snippet generator can attempt to generate a query-independent snippet. Search result 250 of
The representation associated with search result 250 can also be used to present referencing information 252 even if the snippet 251 contained one of the search terms. For example, a search for a person's first and last name may yield a result 250 where the person's first name is used in the page, and thus excerpted and provided as part of snippet 251, and the person's last name is used in a referencing page and is provided as referencing information 252. In such a case, the referencing information 252 can be selected to complement those elements of the search term that may already be present in the page identified as result 250.
An alternative embodiment, illustrated with search result 260, is to provide referencing information 261 comprising multiple anchor texts. As will be described in detail below, a web page may be commonly referenced by one of several different types of anchor text wordings. In such a case, two or more of the most relevant anchor text wordings can be displayed as part of the referencing information 261. Thus, as shown in
Another alternative embodiment is illustrated in connection with search result 270 of
As indicated previously, information obtained from the crawler 42 can be stored in an optimized fashion in the database 50 by the index builder 43. One such optimization can be the selection of select referencing information for the page being added to the database 50. As will be recognized by those skilled in the art, a web page that is part of the WWW may be referenced by thousands, or even millions, of other pages and such referencing is likely to include thousands of different anchor text wordings. Consequently, one embodiment contemplates the selection of select referencing information, such as frequently used anchor text wording, as part of the creation and maintenance of database 50.
Turning to
Once the anchor texts have been sorted according to the wording used in the anchor text, the various pages that use those anchor texts to refer to the page being stored can be identified and categorized accordingly, as shown in step 340. Thus, again using the example of referencing information 261, pages using the anchor text wording “foobar” would be sorted together, while pages using the anchor text wording “total foobar” would likewise be sorted together.
Once the pages have been appropriately split up, a number of factors can be considered to select the anchor text wording that will likely be most relevant. The flow diagram 300 illustrates one embodiment considering factors such as frequency and the static rank of a page. Thus, turning to step 350, for the pages using a particular anchor text wording, the number of times that that anchor text wording is used within one of those pages can be determined. Likewise, at step 360, again for the pages using a particular anchor text wording, the average static rank of those pages can be calculated. As will be known by those skilled in the art, the static rank of a page is a measure of the popularity of that page relative to other pages. And at step 370, again for the pages using a particular anchor text wording, the highest static rank from among those pages can be determined.
One or more of the collected factors, including the frequency determined at step 350, the average static rank determined at step 360, the highest static rank determined at step 370, and any other factors, can be weighed in ranking the anchor text wordings at step 380. In one embodiment, the frequency, average static rank, and highest static rank are each multiplied by a weighting factor that can be adjusted based on empirical evidence and summed. The anchor text wordings can then be sorted according to increasing (or decreasing) total values.
The select referencing information, which is to be stored with the web page to which the above anchor texts have been referring, can be selected from the foremost sorted anchor text wordings. In one embodiment, illustrated by step 390, the top ten anchor text wordings are stored together with the page to which those anchor texts refer. Other embodiments may store a lesser or greater number of anchor text wordings.
In addition to the anchor text itself, additional information can also be stored with the anchor text to enable more efficient determinations of which anchor text to display, as will be described in greater detail below. In one embodiment, along with the anchor text, the database 50 can contain: the frequency with which the anchor text is used among the pages using the anchor text, the highest static rank, minimum static rank and average static rank among the pages using the anchor text, and an encrypted address hash for the page having the highest static rank. As will be recognized by those skilled in the art, page address can be of widely varying lengths. To provide homogenous storage, the addresses can be hashed into values of fixed, pre-determined size. To maintain the security of the hash, the hashed address can be encrypted, and subsequently decrypted once the user has selected an associated element on the results page.
Once stored, the select referencing information can be used to identify the one or more anchor tags that can be presented when referencing information is provided to the user as part of the search results. Turning to
Once the relevant pages have been identified and sorted, the snippet generator 45 can, at step 440, search the database 50 to identify an excerpt from the relevant pages to be displayed along with those pages as search results to the user. At step 450, the snippet generator 45 can verify that the selected snippet contains at least some of the search terms entered by the user. If such a verification can be made, then the front end can generate the search results page, as shown at step 490. In an alternative embodiment, not illustrated in
However, if at step 450, the snippet generator cannot generate a snippet that contains any of the search terms, it can attempt to generate a query-independent snippet at step 460. As explained previously, a page hosting one or more controls may comprise little or no text from which a snippet can be generated. Additionally, a page may be referenced by other pages in such a manner that it appears relevant to the user's search query, but the page itself may not contain any of the terms of the user's query. In such cases, a query-independent snippet, such as the title of the page, or the address of the page, may be generated by the snippet generator at step 470, based upon the determination at step 460. If, at step 460, the snippet generator determines that it cannot generate a query-independent snippet, then it can skip over step 470 and proceed directly to step 480.
As shown in
Turning to
At step 510, anchor texts from among those stored with the web page being proffered as a search result can be parsed for those that share at least one element with the user's query. Such an element can be a word, though it can likewise be any discrete collection of one or more characters, irrespective of linguistic rules. A further similarity that can also be considered in the identification performed at step 510 is the equivalence between any element of the anchor text and the address of the page deemed relevant to the user's query. Step 520 can identify and sort, according to their overall usefulness given the user's query and the text of the page deemed to be a search result, the anchor text wordings that were stored with that page. For example, if the search result page comprises some, but not all, of the terms used in the search query, and one of the stored anchor text wordings comprises the remaining terms of the search query that are not used by the result page, then such an anchor text wording can be deemed more useful and can be ranked accordingly. Usefulness can also comprise similarity, such that the anchor text wordings that were stored with the result page are ordered based on factors including the number of terms they share with the search query, the order of those terms, and the like.
Anchor text wordings which are used by pages having a high static rank can be identified at step 530. At step 540, the previously stored anchor texts can be reviewed to identify those that are in the same language as the user's query. In one embodiment, a pre-check can be made prior to performing step 540 to verify that the user's query is actually in a particular language and not, for example, an alphanumerical query for a serial number. At step 550, anchor text wordings that simply have more terms can be identified, with the expectation that a greater number of terms provides more context for the user.
At step 560, the various identifications performed at steps 510 through 550 can be ranked and assigned weights. For example, the anchor texts with the greatest similarity to the query, as determined at step 520, can be assigned a greater numerical value, while those with less similarity to the query can be assigned a lower numerical value. Subsequently, the assigned numerical value can be multiplied by a weighting value designed to apply a greater weight to those factors that most influence the provision of results that are deemed useful by users. In one embodiment the weighting values assigned to each factor can be continuously revised in response to user feedback.
Once the factors have been assigned values, and have been weighted accordingly, the anchor text to be displayed can be selected at step 570 based on the largest summed values of the weighted factors. The combination of the weighted factors can be termed the “dynamic rank” of the page. In one embodiment, only a single anchor text is selected, while other embodiments contemplate the selection of multiple anchor text wordings for display to the user. For each anchor text wording selected, at step 580, the page with which that anchor text will be associated can be selected based on static rank. Thus, a user selection of the displayed anchor text will direct the user to the page which uses that anchor text to link to the page deemed relevant to the user's query, and which has a higher static rank than all of the other pages that were identified as using the same anchor text wording to also link to the page deemed relevant to the user's query.
As can be seen from the above descriptions, referencing information can be selected for display to the user in order to provide further context for search results for which a snippet with the search terms is not generated. In view of the many possible variations of the subject matter described herein, we claim as our invention all such embodiments as may come within the scope of the following claims and equivalents thereto.
This application claims priority to and is a continuation of co-pending U.S. patent application Ser. No. 11/522,227 entitled “Using Anchor Text to Provide Context” and filed Sep. 15, 2006, which is incorporated herein by reference.
| Number | Date | Country | |
|---|---|---|---|
| Parent | 11522227 | Sep 2006 | US |
| Child | 13886695 | US |
