Challenged Claims of the 'J52 E\ idenct' of Infringement- Coogle ·s Search Engine that uses PagcRank
`Patent
`
`EXHIBIT 2050
`
`26. A non-semantical method
`for numerically representing
`objects in a computer database
`and for computerized searching
`of the numerically represented
`objects in the database, wherein
`direct and indirect relationships
`exist between objects in the
`database, comprising:
`
`Google's search engine that uses PageRank is a non-semantical method for
`numerically representing objects with direct and indirect relationships in a
`database and for computerized searching of the numerically represented
`objects:
`
`Google's search engine that uses PageRank is a non-semantical method for
`numerically representing web pages of the World Wide Web and other similar
`objects that have indirect relationships and for computerize search of the
`numerically represented objects. See
`75 and 2076:
`http://www.google.com/technology/. The method as disclosed in US Patent
`8,631,094 uses a link table representing direct and indirect relationships and nodes
`to process "queries [ ... ] received from client devices [ ... ] such as mobile
`computing devices, mobile communication devices, set-top box television client
`devices, and so on" and does not use or account for words or phrases. See
`'094 Patent abstract; 15:14-19, 42-47; 17:43-20-40.
`
`Web pages have direct and indirect relationships with other objects- through
`'s papers on PageRank recognize
`hyperlink citations between web pages.
`: The Anatomy of a Large-Scale
`that hyperlinks are a form of citation. See
`Hypertextual Web Search Engine at 2.1 ("The citation (link) graph of the web is an
`important resource that has largely gone unused in existing web search engines. We
`have created maps containing as many as 518 million of these hyperlinks. ").
`Indeed, Larry Page, the creator of PageRank algorithm, refers to PageRank as
`"citation ranking" in the title of his work on PageRank. See -
`054: The
`PageRank Citation Ranking: Bringing Order to the Web at 1.
`
`1
`
`

`
`Challenged Claims of the ’352
`Patent
`
`26. A non-semantical method
`for numerically representing
`objects in a computer database
`and for computerized searching
`of the numerically represented
`objects in the database, wherein
`direct and indirect relationships
`exist between objects in the
`database, comprising:
`
`EXHIBIT 2050
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`Google’s search engine that uses PageRank is a non-semantical method for
`numerically representing objects with direct and indirect relationships in a
`database and for computerized searching of the numerically represented
`objects:
`
`Google’s search engine that uses PageRank is a non-semantical method for
`numerically representing web pages of the World Wide Web and other similar
`objects that have indirect relationships and for computerize search of the
`numerically represented objects. See Ex. 2075 and 2076:
`http://www.google.com/technology/. The method as disclosed in US Patent
`8,631,094 uses a link table representing direct and indirect relationships and nodes
`to process “queries […] received from client devices […] such as mobile
`computing devices, mobile communication devices, set-top box television client
`devices, and so on” and does not use or account for words or phrases. See Ex. 2085:
`‘094 Patent abstract; 15:14-19, 42-47; 17:43-20-40.
`
`Web pages have direct and indirect relationships with other objects through
`hyperlink citations between web pages. Google’s papers on PageRank recognize
`that hyperlinks are a form of citation. See Ex. 2053: The Anatomy of a Large-Scale
`Hypertextual Web Search Engine at 2.1 (“The citation (link) graph of the web is an
`important resource that has largely gone unused in existing web search engines. We
`have created maps containing as many as 518 million of these hyperlinks.”).
`Indeed, Larry Page, the creator of PageRank algorithm, refers to PageRank as
`“citation ranking” in the title of his work on PageRank. See Ex. 2054: The
`PageRank Citation Ranking: Bringing Order to the Web at 1.
`
`
`1
`
` 
`
` 
`
`

`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`EXHIBIT 2050
`
`Google’s PageRank patent describes its analysis of the world wide web as a method
`that analyzes a database:
`
`
`“A method assigns importance ranks to nodes in a linked database, such as any
`database of documents containing citations, the world wide web or any other
`hypermedia database.” Ex. 2086: Abstract U.S. Patent No. 6,285,999.
`
`
`PageRank numerically represents web pages and other similar objects that have
`indirect relationships by assigning each web page or object a unique numerical
`identifier:
`
`
`“We convert each URL into a unique integer, and store each hyperlink in a
`database using the integer IDs to identify pages.” Ex. 2054: The PageRank
`Citation Ranking: Bringing Order to the Web at 3.1
`
`
`The purpose of Google’s PageRank algorithm is to facilitate searches of World
`Wide Web and other similar document databases that have indirect relationships
`among the objects within the database. See Ex. 2054: The PageRank Citation
`Ranking: Bringing Order to the Web, abstract (“[w]e show how to apply PageRank
`to search and to user navigation [of web pages]”).
`
`marking objects in the database
`so that each marked object may
`be individually identified by a
`computerized search;
`
`Google’s PageRank algorithm marks objects in the database by assigning a unique
`numerical identifier to each web page or object:
`
`
`“We convert each URL into a unique integer, and store each hyperlink in a
`database using the integer IDs to identify pages.” Ex. 2054: The PageRank
`
`2
`
` 
`
` 
`
`

`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`EXHIBIT 2050
`
`Citation Ranking: Bringing Order to the Web at 3.1
`
`Google’s algorithm assigns unique numerical identifiers to each web page or object
`so objects associated with each assigned identifier may be identified by a
`computerized search:
`
`
`“Additionally, there is a file which is used to convert URLs into docIDs. It is a
`list of URL checksums with their corresponding docIDs and is sorted by
`checksum. In order to find the docID of a particular URL, the URL’s checksum
`is computed and a binary search is performed on the checksums file to find its
`docID. URLs may be converted into docIDs in batch by doing a merge with this
`file. This is the technique the URLresolver uses to turn URLs into docIDs. This
`batch mode of update is crucial because otherwise we must perform one seek for
`every link which assuming one disk would take more than a month for our 322
`million link dataset.” Ex. 2053: The Anatomy of a Large-Scale Hypertextual
`Web Search Engine at 4.2.3.
`
`Google’s PageRank algorithm creates a first numerical representation for each
`object based on the object’s direct relationship with other objects in the database:
`
`Google uses its database of links to create a link matrix for use in the calculation of
`the PageRank algorithm. See Ex. 2053: The Anatomy of a Large-Scale
`Hypertextual Web Search Engine at 2.1.1. Both the database of links and the link
`matrix are numerical representations of direct relationships expressed by
`hyperlinks.
`
`
`3
`
`creating a first numerical
`representation for each
`identified object in the database
`based upon the object's direct
`relationship with other objects
`in the database;
`
` 
`
` 
`
`

`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`EXHIBIT 2050
`
`The database of links is a numerical representation:
`
`
`“It also generates a database of links which are pairs of docIDs. The links
`database is used to compute PageRanks for all the documents.” Ex. 2053: The
`Anatomy of a Large-Scale Hypertextual Web Search Engine at 4.1.
`
`
`
`“Every web page has an associated ID number called a docID which is assigned
`whenever a new URL is parsed out of a web page.” Ex. 2053: The Anatomy of a
`Large-Scale Hypertextual Web Search Engine at 4.1.
`
`PageRank also is generated using a link matrix which is a numerical representation
`of direct relationships:
`
`
`“PageRank or PR(A) can be calculated using a simple iterative algorithm, and
`corresponds to the principal eigenvector of the normalized link matrix of the
`web” Ex. 2053: The Anatomy of a Large-Scale Hypertexual Web Search Engine
`at 2.1.1.
`
`
`This link matrix is further described as:
`
`
`“Stated another way, let A be a square matrix with the rows and column
`corresponding to web pages. Let Au;v = 1=Nu if there is an edge [i.e., link] from
`u to v and Au;v = 0 if not. If we treat R as a vector over web pages, then we have
`R = cAR. So R is an eigenvector of A with eigenvalue c. In fact, we want the
`dominant eigenvector of A. It may be computed by repeatedly applying A to any
`nondegenerate start vector.” See Ex. 2054: The PageRank Citation Ranking:
`4
`
` 
`
` 
`
`

`
`EXHIBIT 2050
`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`Bringing Order to the Web at 4.
`
`
`Thus, Google’s link matrices also constitute a numerical representation of direct
`relationships in the database.
`
`storing the first numerical
`representations for use in
`computerized searching;
`
`The first numerical representation is stored for use in computerized searching:
`
`
`“The indexer performs another important function. It parses out all the links in
`every web page and stores important information about them in an anchors file.
`… The URLresolver reads the anchors file … It also generates a database of
`links which are pairs of docIDs. The links database is used to compute
`PageRanks for all the documents.” Ex. 2053: The Anatomy of a Large-Scale
`Hypertextual Web Search Engine at 4.1 (emphasis added).
`
`
`
`5
`
` 
`
` 
`
`

`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`EXHIBIT 2050
`
`
`Ex. 2053: The Anatomy of a Large-Scale Hypertextual Web Search Engine at 5.2
`tbl.1. As can be seen from the above figure, the links database, which contains
`direct relationships, is clearly stored.
`
`The aforementioned links database is stored, as elaborated upon in Ex. 2054: The
`PageRank Citation Ranking: Bringing Order to the Web, which cites to Ex. 2053:
`The Anatomy of a Large-Scale Hypertextual Web Search Engine (“Details of our
`implementation are in [The Anatomy of a Large-Scale Hypertextual Web Search
`
`6
`
` 
`
` 
`
`

`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`EXHIBIT 2050
`
`analyzing the first numerical
`representations for indirect
`relationships existing between
`or among objects in the
`database;
`
`Engine].”): “Also, all the access to the link database, A, is linear because it is
`sorted. Therefore, A can be kept on disk as well. Although these data structures
`are very large, linear disk access allows each iteration to be completed in about 6
`minutes on a typical workstation.” Ex. 2054: The PageRank Citation Ranking:
`Bringing Order to the Web at 7 (emphasis added).
`
`Furthermore, The PageRank Citation Ranking: Bringing Order to the Web notes,
`“We convert each URL into a unique integer, and store each hyperlink in a database
`using the integer IDs to identify pages.” Ex. 2054: The PageRank Citation
`Ranking: Bringing Order to the Web at 7 (emphasis added).
`
`Google’s PageRank algorithm analyzes the first numerical representations for
`indirect relationships existing between or among objects in the database.
`
`Google obtains and stores information concerning the hyperlink structure of the
`web in a link database. Google’s Software includes methods of analyzing the
`database of links to build a first numerical representation of direct relationships
`between objects (e.g., web page, domain, site, blog, or website). For example, the
`link database is used to create an adapted adjacency matrix for use in the
`calculation of the PageRank algorithm. This algorithm (described in further detail
`below) analyzes direct relationships between objects by building representations
`based on the link structure of the identified objects and by calculating the
`PageRanks of all the objects based on the links database. See limitation directly
`above. These matrices map the direct links between each web page on the Web.
`Ex. 2099: Langville, Amy and Meyer, Carl D, Google's PageRank and Beyond:
`The Science of Search Engine Rankings, at 31-52 (Princeton University Press
`
`7
`
` 
`
` 
`
`

`
`EXHIBIT 2050
`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`2006).
`
`As described in The Anatomy of a Large-Scale Hypertextual Web Search Engine,
`“Google makes heavy use of hypertextual information consisting of link
`structure… [t]he analysis of link structure via PageRank allows Google to evaluate
`the quality of web pages. The use of link text as a description of what the link
`points to helps the search engine return relevant (and to some degree high quality)
`results.” Ex. 2053: The Anatomy of a Large-Scale Hypertextual Web Search
`Engine at 6.2.
`
`Google’s PageRank algorithm is a recursive mathematical algorithm that analyze
`portions of the set of direct links for contributions of websites, web pages, and
`other objects that are indirectly linked to the objects being scored. PageRank
`analyzes the database of links or link matrix of direct relationships for indirect
`relationships “by counting citations or backlinks to a given page. This gives some
`approximation of a page’s importance or quality. PageRank extends this idea by not
`counting links from all pages equally, and by normalizing by the number of links
`on a page.” Ex. 2053: The Anatomy of a Large-Scale Hypertextual Web Search
`Engine at 2.1.1.
`
`
`
`8
`
` 
`
` 
`
`

`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`EXHIBIT 2050
`
`
`Ex. 2053: The Anatomy of a Large-Scale Hypertextual Web Search Engine at 4.1
`fig.1. As can be seen from the above figure, “The links database is used to compute
`PageRanks for all the documents.” Ex. 2053: The Anatomy of a Large-Scale
`Hypertextual Web Search Engine at 4.1. Thus, PageRank is calculated based upon
`
`
`
`9
`
` 
`
` 
`
`

`
`EXHIBIT 2050
`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`a first numerical representation, the Links Database.
`
`The formula for calculating the PageRank of the page, as provided by Ex. 2053:
`The Anatomy of a Large Scale Search Engine, by Brin and Page (1998) is as
`follows:
`
`We assume page A has pages T1....Tn which point to it (i.e., are citations).
`The parameter d is a damping factor which can be set between 0 and 1. We
`usually set d to 0.85. Also C(A) is defined as the number of links going out
`of page A. The PageRank of a page A is given as follows:
`
`PR(A) = (1-d) + d(PR(T1)/C(T1) + .... + PR(Tn)/C(Tn))
`
`The above algorithm, including Google’s subsequent elaborations on the basic
`formula, is applied recursively and measures contributions from indirectly related
`nodes. See Ex. 2099: Langville, Amy and Meyer, Carl D, Google’s PageRank and
`Beyond: The Science of Search Engine Rankings, at 31-52 (Princeton University
`Press 2006).
`
`This formula is an analysis that accounts for indirect relationships in the database
`derived from recursive analysis and weighing of indirect relationships in the
`
`10
`
` 
`
` 
`
`

`
`EXHIBIT 2050
`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`database.
`
`Additionally, “[t]hose skilled in the art will appreciate that this same method can be
`characterized in various different ways that are mathematically equivalent.” Ex.
`2086:’999 Patent, 5:64-66. For example, “the rank of a page A is defined
`according to the present invention as
`
`
`where Bl, ... , Bn are the backlink pages of A, r(Bl), ... , r(Bn ) are their ranks, |B1|, ...
`, |Bn| are their numbers of forward links, and α is a constant in the interval [0,1],
`and N is the total number of pages in the web. Ex. 2086:’999 Patent, 4:15-25.
`Another example is:
`
`
`which states PageRank in terms of the dominant (principal) eigenvector of A. Ex.
`2086:’999 Patent, 5:35-40. Regardless of the precise formula, it is clear that
`PageRank is determined in part by analyzing a first numerical representation of
`direct relationships (e.g., link matrix A) for indirect relationships (e.g., An where n
`≥ 2 or the fact that PR(A) is defined recursively.)
`
`The indirect relationships can be more clearly seen in this figure from Bringing
`
`11
`
` 
`
` 
`
`

`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`EXHIBIT 2050
`
`Order to the Web:
`
`
`
`
`
`
`12
`
` 
`
` 
`
`

`
`EXHIBIT 2050
`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`Ex. 2054: The PageRank Citation Ranking: Bringing Order to the Web at 4 fig.2.
`
`generating a second numerical
`representation of each object
`based on the analysis of the first
`numerical representation;
`
`Google’s PageRank scores constitute a second numerical representation that
`accounts for indirect relationships in the database derived from recursive analysis
`and weighing of indirect relationships in the database. Google uses the links
`database and link matrix to populate values used in the calculation of the PageRank
`algorithm. See quotes above concerning the use of “links database” and “link
`matrix” to calculate PageRank from Ex. 2053: The Anatomy of a Large-Scale
`Hypertextual Web Search Engine at 4.1 and Ex. 2054: The PageRank Citation
`Ranking: Bringing Order to the Web at 4.
`
`13
`
` 
`
` 
`
`

`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`EXHIBIT 2050
`
`
`Ex. 2053: The Anatomy of a Large-Scale Hypertextual Web Search Engine at 4.1
`fig.1.
`
`
`
`As described by the formula disclosed in the previous limitation, PageRank is a
`
`14
`
` 
`
` 
`
`

`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`EXHIBIT 2050
`
`recursive algorithm that analyzes portions of the set of direct links for contributions
`of web pages that are only indirectly related to the selected web page. Thus, the
`algorithm locates indirectly related nodes and scores relationships defined by paths
`of direct links between webpages that are indirectly related to each other. As seen
`in the figure below, the PageRank of the web page represented by node A (also
`referred to as PR(A)) is determined by the PageRank of web links that are back
`linked to A.
`
`Indirect Relationship
`
`
`
`Node A’s PageRank score reflects the scores of the PageRank of Node T1 (PR(T1))
`which is a function of the nodes that back link to it (Node 2), while the PageRank
`of Node 2 is a function of the nodes that back link to it (not shown), and so on and
`so forth. As shown in the arc above, Node A’s PageRank is a function of Node 2’s
`
`15
`
` 
`
` 
`
`

`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`EXHIBIT 2050
`
`PageRank, which is indirectly related to Node A. Node 2’s score is accounted for
`by the algorithm because it is indirectly related to Node A. The PageRank
`algorithm recursively propagates to link lengths up to approximately 50-60
`iterations. See, e.g., Ex. 2054: The PageRank Citation Ranking: Bringing Order to
`the Web at 7 (“As can be seen from the graph in Figure 4 PageRank on a large 322
`million link database converges to a reasonable tolerance in roughly 52
`iterations.”).
`
`Furthermore, Google notes that this formula is applied recursively:
`
`
`“Another intuitive justification is that a page can have a high PageRank if there
`are many pages that point to it, or if there are some pages that point to it and
`have a high PageRank. Intuitively, pages that are well cited from many places
`around the web are worth looking at. Also, pages that have perhaps only one
`citation from something like the Yahoo! homepage are also generally worth
`looking at. If a page was not high quality, or was a broken link, it is quite likely
`that Yahoo's homepage would not link to it. PageRank handles both these cases
`and everything in between by recursively propagating weights through the link
`structure of the web.” Ex. 2053: The Anatomy of a Large-Scale Hypertextual
`Web Search Engine, at 2.1.2.
`
`
`
`The above described process is conducted in a series of iterative and recursive steps
`
`16
`
` 
`
` 
`
`

`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`EXHIBIT 2050
`
`in which the cluster link set is analyzed during each iteration. Each iteration of the
`PageRank algorithm analyzes at least one additional link length of link
`relationships. For example, the second iteration may measure all cluster link
`relationships that are two link lengths from the scored node. Thus, during the
`second iteration, the ranking function (e.g., PageRank(A) will identify and add the
`weight of link relationships two link lengths away from the scored node in a
`manner similar to:
`
`Rank(A) = (1-d) [1+d(contribution of pages of paths of length 1) +
`d2(contribution of pages of paths of length 2)]
`
`The nodes that are identified during the first iteration are of a path length of 1 and
`are directly related to the chosen node. As described below, these nodes are then
`used to locate indirectly related nodes to determine the contribution from those
`nodes.
`
`The third iteration would then measure link relationships that are three links lengths
`from the scoring node in a manner similar to:
`
`Rank(A) = (1-d) [1+d(contribution of pages of paths of length 1) +
`d2(contribution of pages of paths of length 2) + d3 (contribution of pages of
`
`17
`
` 
`
` 
`
`

`
`EXHIBIT 2050
`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`paths of length 3)]
`
`The fourth iteration would measure link relationships that are four links lengths
`from the scoring node in a manner similar to:
`
`Rank(A) = (1-d) [1+d(contribution of pages of paths of length 1) +
`d2(contribution
`
`And so on with each iteration identifying and weighting an additional set of cluster
`link relationships of a given link length.
`
` A
`
` pages “PageRank” is a function of the PageRank of those pages that link to it.
`Likewise, the linked page’s PageRank is also a function of the PageRank of those
`pages that link to it… and so on. Consequently, a PageRank score measures the
`contribution of indirectly related web pages. Consider Ex. 2054: Fig. 3 from The
`PageRank Citation Ranking: Bringing Order to the Web.
`
`
`18
`
` 
`
` 
`
`

`
` 
`
`Challenged Claims of the ’352
`Patent
`
`EXHIBIT 2050
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`
`
`
`As shown above, the indirect relationship from A to B to C is scored and
`accounted for in the algorithm. In determining the PageRank of C, document A
`links to document B and therefore contributes its PageRank (“.2”) (divided
`among its outlinks) to document B. Document B then contributes it PageRank
`(“.2”) (divided by its single outlink) to document C.1 Thus, document C accounts
`for the indirect relationship with document A by the value “.2.” If A was not
`linked to B then document C’s PageRank would not receive a contribution from
`Document A and its score would be different.2
`
`This shows the PageRank values are a second numerical representation generated
`
`                                                            
`1 Document C also receives some pagerank from node A by virtue of its direct link from node A.
`2 It should be further noted that Fig. 3 does not use the damping factor “d” of .85 that is disclosed in the algorithm above. The contribution from document A to document C would
`even more explicitly reflect the indirect relationship because it would be subject to a total damping factor of .72 (.85 from the link from A to B times .85 from the link from
`document B to document C.
`
`
`19
`
` 
`
`

`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`EXHIBIT 2050
`
`using the first numerical representation and accounts for indirect relationships in
`the database.
`
`Furthermore, “[t]hose skilled in the art will appreciate that this same method can be
`characterized in various different ways that are mathematically equivalent.” Ex.
`2086:’999 Patent, 5:64-66. For example, “the rank of a page A is defined
`according to the present invention as
`
`
`where Bl, ... , Bn are the backlink pages of A, r(Bl), ... , r(Bn ) are their ranks, |B1|, ...
`, |Bn| are their numbers of forward links, and α is a constant in the interval [0,1],
`and N is the total number of pages in the web. Ex. 2086:’999 Patent, 4:15-25.
`Another example is:
`
`
`which states PageRank in terms of the dominant (principal) eigenvector of A. Ex.
`2086:’999 Patent, 5:35-40. Regardless of the precise formula, it is clear that
`PageRank is determined in part by analyzing a first numerical representation of
`direct relationships (e.g., link matrix A) for indirect relationships (e.g., An where n
`≥ 2 or the fact that PR(A) is defined recursively.) The analysis of indirect
`relationships is further used to generate the final PageRank score assigned to a
`page, as seen above.
`
`20
`
` 
`
` 
`
`

`
`EXHIBIT 2050
`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`
`
`Google’s PageRank scores are stored and used in connection with its search results.
`
`In Ex. 2053: The Anatomy of a Large-Scale Hypertextual Web Search Engine,
`PageRank values are calculated based upon links and stored for use by a
`“Searcher.”
`
`
`
`storing the second numerical
`representation for use in
`computerized searching; and
`
`21
`
` 
`
` 
`
`

`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`EXHIBIT 2050
`
`
`
`Ex. 2053: The Anatomy of Large-Scale Hypertextual Web Search Engine at
`4.1 fig.1 (red boxes added). “The searcher is run by a web server and uses
`the lexicon built by DumpLexicon together with the inverted index and the
`PageRanks to answer queries.” Ex. 2053: The Anatomy of Large-Scale
`Hypertextual Web Search Engine at 4.1.
`
`“The weights from the current time step are kept in memory, and the previous
`weights are accessed linearly on disk. Also, all the access to the link database,
`A, is linear because it is sorted. Therefore, A can be kept on disk as well.
`Although these data structures are very large, linear disk access allows each
`iteration to be completed in about 6 minutes on a typical workstation.” Ex.
`2054: The PageRank Citation Ranking: Bringing Order to the Web at 7.
`
`PageRank scores are used to search for and identify responsive webpages.
`
`Google’s PageRank score is used to determine if and in what order a particular
`search result will be returned. Since Google only displays the top 10 search results
`on a given page, a page’s pagerank will largely determine whether the page is
`identified for display at all.
`
`The PageRank score of a given webpage is used by Google’s search engine to
`search for objects responsive to an end-user’s query:
`
`
`A major application of PageRank is searching…Google utilizes a number
`of factors to rank search results including standard IR measures, proximity,
`
`22
`
`searching the objects in the
`database using a computer and
`the stored second numerical
`representations, wherein the
`search identifies one or more of
`the objects in the database.
`
` 
`
` 
`
`

`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`EXHIBIT 2050
`
`anchor text (text of links pointing to web pages), and PageRank. While a
`comprehensive user study of the benefits of PageRank is beyond the scope of
`this paper, we have performed some comparative experiments and provide
`some sample results in this paper.
`Ex. 2054: The PageRank Citation Ranking: Bringing Order to the Web,
`Section 5.
`***
`To test the usefulness of PageRank for search we implemented a search engine
`that used only the titles of 16 million web pages. To answer a query, the search
`engine all the web pages whose titles contain all of the query words. Then it
`sorts the results by PageRank. Ex. 2054: The PageRank Citation Ranking:
`Bringing Order to the Web, Section 5-5.1.
`
`***
`
`The citation (link) graph of the web is an important resource that has largely
`gone unused in existing web search engines. We have created maps
`containing as many as 518 million of these hyperlinks, a significant sample of
`total. These maps allow rapid calculation of a web page's
`the
`“PageRank”, an objective measure of its citation importance that
`corresponds well with people's subjective idea of importance. Because of
`this correspondence, PageRank is an excellent way to prioritize the
`results of web keyword searches. For most popular subjects, a simple text
`matching search that is restricted to web page titles performs admirably when
`PageRank prioritizes the results (demo available at google.stanford.edu). For
`the type of full text searches in the main Google system, PageRank also helps
`23
`
` 
`
` 
`
`

`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`EXHIBIT 2050
`
`
`
`
`
`
`
`a great deal.
`Ex. 2053: The Anatomy of Large-Scale Hypertextual Web Search Engine,
`Section 2.1.
`
`“The searcher is run by a web server and uses the lexicon built by
`DumpLexicon together with the inverted index and the PageRanks to answer
`queries.”
`
`Ex. 2053: The Anatomy of Large-Scale Hypertextual Web Search Engine,
`Section 4.1, emphasis added
`
`“As an example which illustrates the use of PageRank, anchor text, and
`proximity, Figure 4 shows Google’s results for a search on ‘bill clinton’.”
`Ex. 2053: The Anatomy of Large-Scale Hypertextual Web Search Engine at 5.
`
`24
`
` 
`
` 
`
`

`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`EXHIBIT 2050
`
`
`
`
`Ex. 2053: The Anatomy of Large-Scale Hypertextual Web Search Engine at 5
`fig.4.
`
`Another important application and embodiment of the present invention is
`directed to enhancing the quality of results from web search engines. In this
`application of the present invention, a ranking method according to the
`invention is integrated into a web search engine to produce results far
`
`25
`
` 
`
` 
`
`

`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`EXHIBIT 2050
`
`superior to existing methods in quality and performance. …
`
`Once a set of documents is identified that match the search terms, the list of
`documents is then sorted with high ranking documents first and low ranking
`documents last. U.S. Patent No. 6,285,999, Method for Node Ranking in a
`Linked Database, 8:6-11, 42-46.
`
`Search results including objects are displayed to the user. The user may click
`on a link and request Google to display the full text of the object. The Software
`provides an interface for receiving the inputs, processes the inputs and sends
`appropriate instructions. The Software also automatically displays a portion of
`the object and in some cases the full text of the object.
`
`
`26
`
` 
`
` 
`
`

`
`Challenged Claims of the ’352
`Patent
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`EXHIBIT 2050
`
`
`Google Search for “Google PageRank”
`
`
`
`
`
`
`
`
`
`
`
`32. The non-semantical method
`of claim 26, wherein the step of
`analyzing further comprises the
`
`The score of relationships of nodes indirectly related to the selected node are each
`uniquely based upon different factors so that some are weighed more than others.
`PageRank inventors Lawrence Page and Sergey Brin noted that the PageRank
`
`27
`
` 
`
` 
`
`

`
`Challenged Claims of the ’352
`Patent
`
`step of weighing, wherein some
`indirect relationships are
`weighed more heavily than
`other indirect relationships.
`
`EXHIBIT 2050
`
`Evidence of Infringement – Google’s Search Engine that uses PageRank
`
`calculation naturally weighs some indirect relationships more heavily than others:
`
`
`
`“Anot