United States Patent & Trademark Office
`Patent Trial & Appeal Board
`_________________________________________
`
`IRON DOME LLC
`Petitioner
`
`v.
`
`CHINOOK LICENSING DE LLC
`Patent Owner
`_________________________________________
`
`Petition for Inter Partes Review
`of
`
`Patent No. 7,047,482 (to Gary Odom)
`Titled: Automatic directory supplementation
`Issue date: May 16, 2006
`_________________________________________
`
`
`
`
`
`
`
`
`
`
`For Paralegal:
`Number of Claims Challenged = 19
`Power of Attorney enclosed
`Fee paid online by credit card
`
`Contact: Steven Yu
`Phone: 202.262.0426
`Email: syu@patent-intercept.com
`
`

`
`US 7,047,482
`
`Table of Contents
`
`
`I.
`
`II.
`
`Introductory Matters .............................................................................. - 1 -
`A.
`Relief Requested ............................................................................................. - 1 -
`B.
`Grounds for Standing .................................................................................... - 1 -
`C. Mandatory Notices ......................................................................................... - 1 -
`Prior Art References ............................................................................... - 3 -
`A.
`The claims have an effective filing date of February 28, 2001 .............. - 3 -
`B.
`List of Prior Art .............................................................................................. - 3 -
`III. Technical Background & Claim Construction...................................... - 4 -
`A.
`Technical Background of the Challenged Patent ..................................... - 4 -
`B.
`Claim Construction ........................................................................................ - 5 -
`IV. Grounds for Challenge ........................................................................... - 9 -
`A.
`Chen - primary prior art reference .............................................................. - 9 -
`B.
`Lieberman - second prior art reference ...................................................... - 9 -
`Claim Analysis .......................................................................................- 10 -
`Independent Claim 1 ................................................................................... - 10 -
`Claim 2............................................................................................................ - 15 -
`Claim 3............................................................................................................ - 15 -
`Claim 4............................................................................................................ - 16 -
`Claim 5............................................................................................................ - 16 -
`Claim 6............................................................................................................ - 17 -
`Claim 7............................................................................................................ - 17 -
`Claim 9............................................................................................................ - 18 -
`Claim 10 ......................................................................................................... - 18 -
`Independent Claim 11 ................................................................................. - 19 -
`Claim 12 ......................................................................................................... - 23 -
`Claim 13 ......................................................................................................... - 23 -
`Claim 14 ......................................................................................................... - 24 -
`ii
`
`V.
`
`
`
`

`
`US 7,047,482
`
`Claim 15 ......................................................................................................... - 24 -
`Independent Claim 16 ................................................................................. - 24 -
`Claim 17 ......................................................................................................... - 28 -
`Claim 18 ......................................................................................................... - 29 -
`Claim 19 ......................................................................................................... - 29 -
`Claim 20 ......................................................................................................... - 29 -
`
`
`
`iii
`
`

`
`US 7,047,482
`
`Exhibit List
`Exh. 1001 U.S. Patent No. 7,047,482 (‘challenged patent’)
`
`Exh. 1002 Liren Chen & Katia Sycara, “WebMate: A Personal Agent for
`Browsing and Searching” in Proceedings of the Second International
`Conference on Autonomous Agents. Sponsored by ACM SIGART in
`Minneapolis/St. Paul, MN; May 9-13, 1998. Selected pages: Table
`of Contents, pp. 132-139 (‘Chen’)
`
`Exh. 1003 Henry Lieberman, “Letizia: An Agent That Assists Web
`Browsing” in Proceedings of the Fourteenth International Joint Conference
`on Artificial Intelligence. Sponsored by IJCAII in Montréal, Québec,
`Canada; August 20-25, 1995. Selected pages: Table of Contents,
`pp. 924-929 (‘Lieberman’)
`
`Exh. 1004 “BRIEF FOR APPELLANT” dated Feb. 18, 2005 in the prosecution
`history of the challenged patent (‘Appeal Brief’)
`
`
`
`
`Citation Form Used
`
`
`Reference to supporting documents indicated by “@____”.
`Citations to U.S. Patents are shown as [column number : line numbers].
`Citations to line-numbered documents are shown as [page number : line numbers].
`Claim terms are distinguished from other text by “underlining.”
`
`
`
`
`
`iv
`
`

`
`US 7,047,482
`
`Issue Presented
`
`
`
`The challenged patent was recently asserted in patent infringement lawsuits
`
`against Facebook, Scribd, Hulu, Pandora, Match.com, and others for operating
`
`websites that observe a user’s selections and make personalized recommendations for
`
`other similar selections (e.g. “other music you may like”). A search of the prior art
`
`reveals that the claims of the challenged patent are obvious over a Carnegie Mellon
`
`University graduate student’s published work on web browsing agents that explore the
`
`web and recommend webpages of interest to the user.
`
`
`
`v
`
`
`
`
`
`

`
`US 7,047,482
`
`I.
`
`Introductory Matters
`IRON DOME LLC (‘Petitioner’) petitions for Inter Partes Review (‘IPR’) of U.S.
`
`Patent No. 7,047,482 (‘challenged patent’ @Exh. 1001), which is owned by CHINOOK
`
`LICENSING DE LLC.
`
`A. Relief Requested
`Petitioner requests cancellation of claims 1-7 and 9-20 (total of 19 claims) of
`
`the challenged patent for obviousness under 35 U.S.C. § 103.
`
`B. Grounds for Standing
`Petitioner certifies that the challenged patent is available for IPR and that
`
`Petitioner is not barred or estopped from requesting an IPR challenging the patent
`
`claims on the grounds identified in this petition.
`
`C. Mandatory Notices
`Real Parties-in-Interest: (1) IRON DOME LLC, a Virginia limited liability
`
`
`
`company, which is a wholly-owned subsidiary of ROZMED LLC, a Virginia limited
`
`liability company; and (2) Steven S. Yu, M.D., an individual residing in Rockville,
`
`Maryland and the managing member of ROZMED LLC.
`
`
`
`Individual Steven S. Yu, M.D. declares that there are no other parties that are
`
`funding this IPR, nor participating in any manner in this IPR; and further that this
`
`statement is being made with the knowledge that willful false statements and the like
`
`so made are punishable by fine or imprisonment, or both, under Section 1001 of Title
`
`18 of the United States Code.
`
`
`
`
`- 1 -
`
`

`
`US 7,047,482
`
`
`
`Related Matters: The challenged patent has been asserted by the patent owner
`
`in litigation against several defendants alleging infringement by websites that observe a
`
`subscriber’s selections and recommend other similar selections to the subscriber. On
`
`or about December 20, 2013 and January 20, 2014, the patent owner Chinook
`
`Licensing DE LLC filed civil actions 1:13-cv-02077 through 02079, 1:14-cv-00073
`
`through 00077, and 1:14-cv-00105 in the U.S. District Court for Delaware.
`
`
`
`Individual Steven S. Yu, M.D. declares that Petitioner is not a party to any of
`
`these civil actions, nor has Petitioner been given or taken any direct financial interest
`
`relating to the outcome of these civil actions; and further that this statement is being
`
`made with the knowledge that willful false statements and the like so made are
`
`punishable by fine or imprisonment, or both, under Section 1001 of Title 18 of the
`
`United States Code.
`
`
`
`Electronic Service: Petitioner consents to service by email at:
`
`syu@patent-intercept.com.
`
`Lead Counsel
`
`Back-Up Counsel
`
`Steven Yu (Reg. No. 58,776)
`ROZMED LLC
`PO Box 10034
`Gaithersburg, MD 20898
`Tel: 202.262.0426
`Email: syu@patent-intercept.com
`
`John J. Yim (Reg. No. 47,197)
`JOHN J. YIM & ASSOCIATES LLC
`7600 Leesburg Pike
`East Building, Suite 470
`Tysons Corner, VA 22043
`Tel: 703.749.0500
`Fax: 202.379.1723
`Email: jyim@yimassociates.com
`
`
`- 2 -
`
`
`
`
`
`

`
`US 7,047,482
`
`II. Prior Art References
`A. The claims have an effective filing date of February 28, 2001
`The challenged patent was granted from application Serial No. 09/796,235 filed
`
`on February 28, 2001. This application does not claim priority to any prior-filed
`
`applications. Accordingly, the earliest possible effective filing date for the claims of
`
`the challenged patent is February 28, 2001.1
`
`B. List of Prior Art
`The prior art publications referenced herein are as follows.
`
`1. Liren Chen & Katia Sycara, “WebMate: A Personal Agent for Browsing
`and Searching” in Proceedings of the Second International Conference on
`Autonomous Agents. Sponsored by ACM SIGART in Minneapolis/St.
`Paul, MN; May 9-13, 1998. Selected pages: Table of Contents, pp. 132-
`139 (‘Chen’; Exh. 1002)2
`
`2. Henry Lieberman, “Letizia: An Agent That Assists Web Browsing” in
`Proceedings of the Fourteenth International Joint Conference on Artificial Intelligence.
`Sponsored by IJCAII in Montréal, Québec, Canada; August 20-25, 1995.
`Selected pages: Table of Contents, pp. 924-929 (‘Lieberman’; Exh.
`1003)3
`
`
`1 We reserve the right to dispute whether the challenged claims should legitimately
`have the benefit of this filing date (i.e. lack of written support).
`
` 2
`
` Chen is an article published in a book distributed to participants at a symposium
`proceeding in May 1998. The copy held by the Library of Congress in Washington,
`D.C. is date stamped October 18, 1999. Therefore, Chen is prior art under 35 U.S.C §
`102(b).
`3 Lieberman is an article published in a book distributed to participants at a
`symposium proceeding in August 1995. The copy held by the Georgetown University
`Libraries in Washington, D.C. is date stamped March 25, 1998. Therefore, Lieberman
`is prior art under 35 U.S.C § 102(b).
`
`
`
`
`- 3 -
`
`

`
`US 7,047,482
`
`Neither of the above publications were cited in the original prosecution of the
`
`challenged patent.
`
`
`III. Technical Background & Claim Construction
`A. Technical Background of the Challenged Patent
`
`The challenged patent (Exh. 1001) describes a computer software program that
`
`assists a user in browsing the Internet by making recommendations for webpages that
`
`might interest the user (sometimes referred to as a browsing agent). To make these
`
`recommendations, the software examines a set of items that the user has selected (e.g.
`
`the user’s bookmark folder of “favorite” websites) as indicative of the user’s field of
`
`interest.
`
`FIG. 2 of the challenged patent shows a directory 3 of links or documents (e.g.
`
`webpages) that the user has selected and categorized according to a topic as
`
`designated by the directory title 5. @3:14-18. The browsing agent applies linguistic
`
`analysis techniques to the textual content of the selected items and extracts words that
`
`appear to be relevant to the user’s field of interest (e.g. by analysis of word frequency,
`
`word placement, syntax, etc.). @3:33-36. Extracted keywords are rated or ranked
`
`according to such factors as location in the document, prominence, and frequency of
`
`appearance. @4:22-33.
`
`This keyword extraction and ranking process is performed on all the items in
`
`directory 3 and those that best represent the content of the directory 3 as a whole are
`
`
`
`
`- 4 -
`
`

`
`US 7,047,482
`
`selected as directory keywords 88. @4:63-5:17. Using the directory keywords 88, the
`
`browsing agent searches the network for other items having textual content similar to
`
`those in directory 3. @5:18-23. The user can set a breadth threshold to widen or
`
`narrow the scope of the search. @4:54-62. As shown in FIG. 6, the original directory
`
`3K (the user’s interest in the musical group King Crimson) is supplemented with the
`
`search results in a new box beneath (directory supplementation 6K). The new links 1F
`
`are displayed along with a ranking 66 of their relevance. @5:64-67.
`
`B. Claim Construction
`
`In the context of an inter partes review, claim terms must be given their broadest
`
`reasonable interpretation in view of the specification.
`
`1. “autonomously” (see claims 10 and 11)
`
`This term relates to how the software operates without direct user
`
`participation. This term was added to the claims by amendment during the
`
`prosecution of the challenged patent, but it does not appear anywhere in the
`
`specification of the challenged patent. However, the challenged patent explains the
`
`“automatic” operation of the software agent as follows:
`
`Directory supplementation 6 may be enabled 10 by default, by software-
`determined protocol, or by user determination. Automatically
`supplementing a directory 6 refers to adding links 1 or documents 2 to a
`directory 3 without a user having to search 12 or manually add links 1 to
`that directory 3.
`
`
`
`
`- 5 -
`
`

`
`US 7,047,482
`
`@4:35-40. Thus, the term “autonomously” is synonymous with “automatically” and
`
`means that the directory supplementation process occurs “without a user having to
`
`search [] or manually add links [] to that directory [].”
`
`2. “without contemporaneous user input” (and variations thereof)
`
`This is another term relating to how the software operates without direct user
`
`participation. This term was added to the claims by amendment during the
`
`prosecution of the challenged patent, but it does not appear anywhere in the
`
`specification of the challenged patent. We therefore turn to the applicant’s statements
`
`during the prosecution of the challenged patent to understand the meaning of this
`
`term. The applicant gave the following explanation in his Appeal Brief:
`
`So, Examiner … tacitly concurred with appellant, that, in context, the two
`limitations applicable to the meaning of ‘without user input’ comprise:
`
`1. no user input of search parameters;
`2. no user input of search locations.
`
`That is exactly what appellant had explained in his 08/27/2004 reply to the
`first office action rejection.
`
`@Appeal Brief 14 top (Exh. 1004). Thus, according to the applicant’s statements in
`
`the prosecution history, the term “without contemporaneous user input” means that
`
`the software works with “no user input of search parameters” and “no user input of
`
`search locations.”
`
`
`
`
`- 6 -
`
`

`
`US 7,047,482
`
`3. “searching as a background operation” (and variations thereof)
`
`This term relates to how the software runs as a background process relative to
`
`other processes that the computer may be running. This term was added to the claims
`
`by amendment during the prosecution of the challenged patent, but it does not appear
`
`anywhere in the specification of the challenged patent. We therefore turn to the
`
`applicant’s statements during the prosecution of the challenged patent to understand
`
`the meaning of this term. The applicant gave the following explanation in his Appeal
`
`Brief:
`
` [The software process in my patent application] is fairly characterized as
`lazy because time is not of the essence. A user doesn’t initiate search: the
`process works in the background, without arousing expectation of quick
`results.
`
`@Appeal Brief 2 ¶2nd (Exh. 1004). Thus, according to the applicant’s statements in
`
`the prosecution history, the term “searching as a background operation” means that
`
`the searching operation occurs in the background, simultaneously but with a lower
`
`priority to other operations that the computer may be performing.
`
`4. “precondition” (claims 10 and 11)
`
`This term relates to the selections made by the user before the software begins
`
`its autonomous operation. This term was added to the claims by amendment during
`
`the prosecution of the challenged patent, but it does not appear anywhere in the
`
`specification of the challenged patent. We therefore turn to the applicant’s statements
`
`
`
`
`- 7 -
`
`

`
`during the prosecution of the challenged patent to understand the meaning of this
`
`term. The applicant gave the following summary of his invention in his Appeal Brief:
`
`US 7,047,482
`
` [My patent application] describes an autonomous search mechanism,
`solving the problem of finding similar documents to ones already known
`without any user effort whatsoever. The only precondition to initiating the
`claimed process is user placement of one or more documents in a file
`system directory as reference material for guiding the search.
`
`…
`
`As an exemplary use-case scenario, a user browses the web, saving topically-
`related document links in the same web-favorites folder. Once this
`precondition is met, the claimed invention software kicks in: deriving
`keywords from the saved documents, thus discerning the topic of interest,
`then searching for other related documents, resulting in supplementing the
`directory with newly-found documents - hence the title of [my patent
`application]: ‘automatic directory supplementation’.
`
`@Appeal Brief 2 top (italics added; Exh. 1004). Further clarification of the term
`
`“precondition” is given by the manner in which the applicant distinguished his
`
`invention over the prior art:
`
`By contrast, the claimed invention relies solely upon documents in a
`directory, without relying upon user input. Yes, [in my invention] a user
`must first put the documents in the directory, but that is a precondition; user
`input is not required for the claimed process to work, unlike [the prior art].
`
`@Appeal Brief 14 ¶6th (italics added).
`
`Thus, according to the applicant’s statements in the prosecution history, the
`
`
`
`
`- 8 -
`
`

`
`US 7,047,482
`
`term “precondition” means the selection of items that the user has identified as being
`
`of interest and encompasses at least “user placement of one or more documents in a
`
`file system directory as reference material for guiding the search” such as “saving
`
`topically-related document links in the same web-favorites folder.” In the words of
`
`the applicant, when the user has performed this step, “this precondition is met.”
`
`
`IV. Grounds for Challenge
`Petitioner requests cancellation of claims 1-7 and 9-20 of the challenged patent.
`
`Claims 1-7 and 9-19 are obvious over Chen alone. Claim 20 is obvious over Chen in
`
`view of Lieberman. Claim 8 is not of commercial interest to Petitioner and is
`
`therefore omitted from this IPR.
`
`A. Chen - primary prior art reference
`Chen (Exh. 1002) is an article published in 1998 and describes the WebMate
`
`software, which is a web browsing agent that automatically learns the user’s interest
`
`and finds webpages that match the user’s interest. @132 L Abstract. WebMate
`
`compiles a “personal newspaper” of the recommended webpages. @Id.
`
`B. Lieberman - second prior art reference
`Lieberman (Exh. 1003) is an article published in 1995 and describes the Letizia
`
`software, which is a “user interface agent that assists a user browsing the World Wide
`
`Web.” @924 L Abstract. The software agent infers the user’s interest from browsing
`
`behavior and explores other items that may interest the user. @Id. Chen specifically
`
`
`
`
`- 9 -
`
`

`
`US 7,047,482
`
`cites to Lieberman as “Related work.” @Chen 138 R, #17 (Exh. 1002). Thus, it is
`
`appropriate to combine Chen with Lieberman for an obviousness analysis.
`
`
`
`V. Claim Analysis
`The challenged patent has three independent claims, which are claims 1, 11,
`
`and 16, all being for a “computer-implemented method,” i.e. computer software. In
`
`general, all three independent claims are directed to a software agent that
`
`automatically (“without contemporaneous user input”) finds items (e.g. links or
`
`webpages) that match the user’s interest and adds them to a directory. A well-known
`
`example would be a web browsing agent that analyzes a user’s bookmark folder of
`
`“favorite” links, searches the Internet for links with similar content, and recommends
`
`other links that might be of interest to the user.
`
`Independent Claim 1
`
`a) (preamble) “A computer-implemented method for augmenting a
`directory without contemporaneous user input comprising:”
`
`Chen (Exh. 1002) describes the WebMate software agent, which automatically
`
`learns a user’s interest by extracting keywords from documents selected by the user
`
`and “automatically spiding news sources” (“spiding” means crawling the web, as a
`
`spider would) on the Internet for webpages having similar content. @Abstract. As
`
`explained above, the term “without contemporaneous user input” means that the
`
`software works with “no user input of search parameters” and “no user input of
`
`
`
`
`- 10 -
`
`

`
`US 7,047,482
`
`search locations.” @IPR 6. More detailed explanation about this and other features of
`
`WebMate are given below.
`
`b) (claim 1) “accessing at least a first document via a first directory
`without contemporaneous user selection of said first document, said
`first document comprising at least in part topical textual content”
`This claim term is essentially stating that the software agent automatically
`
`examines the text in the documents selected by the user (e.g. bookmarked links in a
`
`“favorites” folder). In the challenged patent, an example of a “first directory” is given
`
`in FIG. 6. @Exh. 1001 p. 6. Here, the directory 3K is titled “King Crimson” and
`
`contains links related to this particular musical group. The directory 3K (upper
`
`portion) corresponds to the “first directory.” As explained above, the software agent
`
`analyzes the textual content of the linked pages in this directory for keywords that
`
`appear to represent the user’s interest. Regarding the claim term “without
`
`contemporaneous user selection,” the links in directory 3K were previously selected
`
`by the user, and the software agent automatically accesses the linked pages in directory
`
`3K.
`
`WebMate works in a similar manner by “learning user interests incrementally
`
`and with continuous update and automatically providing documents (e.g. a
`
`personalized newspaper) that match the user interests.” @132 R ¶bott. WebMate
`
`develops the user profile automatically and “unobtrusively” by keyword extraction
`
`from documents that the user selects by marking them as “I like it.” @132 L Abstract,
`
`134 L mid. The “user can provide multiple pages as similarity/relevance guidance for
`
`- 11 -
`
`
`
`

`
`US 7,047,482
`
`the search.” @132 L Abstract. Thus, WebMate accesses the user-selected documents
`
`“without contemporaneous user selection,” which is understood to mean “no user
`
`input of search parameters” and “no user input of search locations.” @IPR 6.
`
`“In order to save on storage space,” WebMate “doesn’t keep any of the
`
`previous positive example documents.” @133 R ¶top. However, storage space is not a
`
`concern for higher capacity machines and it would be obvious to keep the set of “I
`
`like it” documents selected by the user as a collection of links instead of discarding
`
`them. This collection of “I like it” documents selected by the user would constitute a
`
`“first directory” recited in the claim. To be explained in detail below, these user-
`
`selected documents have “topical textual content.”
`
`c) (claim 1) “deriving at least one keyword indicative of at least one
`topical content from said first document”
`
`
`
`Once the user has selected the items of interest, WebMate automatically parses
`
`the selected web documents and extracts words contained therein, including those in
`
`the title and headers. @133 R – 134 L. Using the extracted words, WebMate then
`
`constructs a word vector for the document. @133 R ¶2nd. The word vector is an
`
`aggregate representation of the content as a series of word elements, each weighted
`
`according to its frequency of occurrence in the document. @Id. The combined word
`
`vectors become the user profile. @Id.
`
`
`
`
`- 12 -
`
`

`
`US 7,047,482
`
`d) (claim 1) “searching as a background operation a plurality of
`documents in storage in at least one computer without
`contemporaneous user input of a search location, such that said
`search comprises searching for documents related by said at least one
`keyword to said first document, thereby accessing a second
`document”
`
`
`
`This claim term is essentially stating that the search for similar documents on
`
`the Internet or network occurs as a background process on the computer. With the
`
`user’s interest profile created, WebMate can search the entire web for pages relevant
`
`to the user’s interest, or monitor only a selection of websites designated by the user.
`
`@134 §3.2. If the user has selected a particular list of websites (URLs) for monitoring,
`
`WebMate can search these websites to find relevant webpages based on their
`
`similarity to the word vector in the user’s profile. @134 L ¶bott. If the user does not
`
`provide a list of websites to monitor, then WebMate can simply construct a search
`
`query “using the top several words in the current [user] profile and sends it to search
`
`engines (e.g. Altavista, Yahoo).” @134 R ¶2nd.
`
`
`
`When WebMate is searching selected websites for pages that match the word
`
`vector in the user’s profile, or searching the entire web using the top words in the
`
`user’s profile, it is “searching for documents related by said at least one keyword to
`
`said first document.” The webpages being examined by WebMate constitute a
`
`“second document” which are “in storage in at least one computer,” i.e. in the
`
`computers running the websites.
`
`
`
`
`- 13 -
`
`

`
`US 7,047,482
`
`
`
`WebMate can perform its search “in the middle of the night when the network
`
`traffic is low” such that “[i]n the morning, the user can read the recommended
`
`personal newspaper.” @134 R top. WebMate can also perform its search
`
`contemporaneously “[i]f the result is needed immediately.” @Id. R ¶ 2nd. As “Related
`
`work,” Chen also points to the Letizia web browsing agent which “can recommend
`
`nearby pages by doing lookahead search.” @138 R top. Thus, Chen discloses three
`
`different timeframes in which searching can be performed: (1) overnight during
`
`machine idle time, (2) immediately, and (3) as a “lookahead search” while the user is
`
`browsing the Internet. All three searching methods are performed “without
`
`contemporaneous user input of a search location.” Moreover, from these teachings
`
`about the different timeframes for performing the search, it would be obvious to have
`
`WebMate perform “searching as a background operation” in the same manner as
`
`Letizia’s “lookahead search.”
`
`e) (claim 1) “determining relevance of said second document to said at
`least one keyword”
`
`
`
`WebMate decides whether to recommend the pages it has found to the user on
`
`the basis of whether the page content “similarity is greater than some threshold” as
`
`compared against the user’s profile (which is represented by word vectors). @134 L
`
`¶bott. (see also R ¶2nd).
`
`
`
`
`- 14 -
`
`

`
`US 7,047,482
`
`f) (claim 1) “adding a reference to said second document in a results
`directory.”
`
`
`
`If the user has selected websites for monitoring, WebMate can perform the
`
`search overnight and in the morning, provide the user with a “personal newspaper”
`
`listing the recommended pages. @134 L ¶bott. If WebMate performs a web-wide
`
`search, WebMate analyzes the search results and presents those meeting the required
`
`threshold for similarity “in descending order of relevance.” @134 R ¶2nd. In both of
`
`the aforementioned situations, WebMate works by “adding a reference to said second
`
`document [i.e. the recommended webpage] in a results directory.”
`
`Claim 2
`
`Claim 2 depends from claim 1 and specifies that “at least part of said storage is
`
`on a different computer than the computer storing said first directory.” WebMate
`
`performs a search of the World Wide Web for documents that are relevant to the
`
`user’s interest. The World Wide Web is a network of many different computers
`
`throughout the world. For example, Chen describes an experiment with WebMate in
`
`which 14 news sites covering technology news were monitored by WebMate for news
`
`articles matching the user’s interest. @134 R ¶3rd. The 14 news sites on the Internet
`
`selected for monitoring are on a “different computer” than the user’s computer. @Id.
`
`Claim 3
`
`Claim 3 depends from claim 1 and adds the step of “deriving a plurality of
`
`keywords.” As explained above, WebMate parses through the documents selected by
`
`- 15 -
`
`
`
`

`
`US 7,047,482
`
`the user (marked as “I like it”) and extracts words from the documents to generate a
`
`weighted word vector representing the user’s field of interest.
`
`Claim 4
`
`Claim 4 depends from claim 3 and adds the step of “ranking at least two of said
`
`plurality of keywords.” As explained above for claim 1(c), WebMate generates a word
`
`vector for the words in the document, and each word in the vector is weighted
`
`according to its frequency of occurrence. @IPR 12. In addition, WebMate can
`
`operate by constructing a search query “using the top several words in the current
`
`[user] profile and sends it to popular search engines (e.g. Altavista, Yahoo).” @134 R
`
`¶2nd. In Chen’s experimental demonstration of WebMate, the “top 5 words” were
`
`used to construct a search query for the Lycos search engine. @138 L ¶3rd. Both of
`
`the aforementioned functions, weighting of the words in the word vector or selection
`
`of the “top several words,” constitutes “ranking at least two of said plurality of
`
`keywords.”
`
`Claim 5
`
`
`
`Claim 5 depends from claim 1 and adds the step of “accessing a plurality of
`
`documents in said first directory.” WebMate performs the parsing routine “[f]or each
`
`positive example (i.e. an HTML documents [sic] that the user has marked ‘I like It’).”
`
`@133 R ¶5th. Thus, if there are multiple web documents marked by the user as “I like
`
`
`
`
`- 16 -
`
`

`
`it,” then WebMate will perform the parsing routine on each of the multiple web
`
`US 7,047,482
`
`documents.
`
`Claim 6
`
`
`
`Claim 6 depends from claim 1 and adds the step of “signifying the relevance of
`
`said second document to documents in the first directory when displaying said results
`
`directory.” In the challenged patent, FIG. 6 shows the results directory 6K with the
`
`new links 1F displayed along with a ranking 66 of their relevance. @Exhibit 1001 p. 6
`
`(see also 5:64-67).
`
`
`
`In WebMate, the “personal newspaper” provided to the user “sorts all the
`
`recommended pages in decreasing order of similarity.” @134 L ¶bott. Alternatively, if
`
`instead WebMate performs a search query, WebMate lists the recommended pages in
`
`“descending order of relevance.” @Id. R ¶2nd. Moreover, Chen shows the search
`
`results returned by the Lycos search engine in one of the experimental demonstrations
`
`of WebMate. @138 L ¶3rd. Regarding the Lycos search results shown, the “content of
`
`links marked with ‘*’ are similar to the content of the page given as the ‘relevant’
`
`feedback.” @Id. Thus, WebMate performs the step of “signifying the relevance of
`
`said second document … when displaying said results directory.”
`
`Claim 7
`
`Claim 7 depends from claim 1 and adds the step of “comparing the relevance
`
`of said second document to a preset threshold.” WebMate recommends a webpage to
`
`
`
`
`- 17 -
`
`

`
`the user if its “similarity is greater than some threshold.” @134 L ¶bott. (see also R
`
`US 7,047,482
`
`¶2nd).
`
`Claim 9
`
`Claim 9 depends from claim 1 and adds the step of “displaying said results
`
`directory.” As explained above for claim 1(f), WebMate can provide the user with a
`
`“personal newspaper” listing the recommended pages or present the search results “in
`
`descending order of relevance.” @IPR 15. This constitutes “displaying said results
`
`directory.”
`
`Claim 10
`
`Claim 10 depends from claim 1 and adds the step of “recognizing a
`
`preco