DETAILED ACTION
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1 to 2, 7 to 11, and 16 to 17 are rejected under 35 U.S.C. 103 as being unpatentable over Tiwari (U.S. Patent Publication 2019/0102374) in view of Consul et al. (U.S. Patent No. 9,165,056).
Concerning independent claims 1 and 10, Tiwari discloses a method and system for predicting future trending topics, comprising:
“reading data from at least one data file including a plurality of textual data entries” – text extractor 344 can receive through interface 342 a set of posts or content items (“a plurality of textual data entries”); text extractor 344 can extract text from the received set of posts by taking textual content from a post (¶[0032]: Figure 3); process 400 can extract text from a current post (¶[0042]: Figure 4: Step 410);
“processing the plurality of textual data entries including one or more of removing word cases and punctuation, lemmatizing nouns, removing a first category of stop words, and removing a second category of stop words including thematic or data file specific stop words and phrases” – n-gram generator 348 can normalize the extracted text for each received post; n-gram generator 348 can remove from a cumulative set n-grams that contain certain specified stop words (“processing the plurality of textual data entries including one or more of . . . removing a first category of stop words . . .”); a set of stop words can be words that appear above a threshold frequency in a language, words determined to be offensive, or manually selected words, e.g., words determined to be unhelpful for determining a trending topic (“removing a second category of stop words including thematic or data file specific stop words and phrases”); stop words can include numbers (¶[0034] - ¶[0035]: Figure 3); process 400 can remove n-grams from a cumulative set of n-grams that include one or more stop words; a defined set of stop words can include any words that appear above a threshold frequency in a language, words determined to be offensive, or manually selected words, e.g., words determined to be unhelpful for determining a trending topic; stop words can include numbers (¶[0047]: Figure 4: Step 422);
“applying natural language processing (NLP) to each of the processed plurality of textual data entries including: identifying word level n-grams for one or more selectable word level n-gram lengths” – text is extracted from each post, text is normalized and tokenized, and tokens are organized into n-grams; in various implementations, n-grams can be limited to an exact number of words, e.g., two (¶[0010]); n-gram generator 348 can normalize text, and organize the tokenized text into n-grams of a particular length (¶[0034]: Figure 3); n-grams can be tri-grams (¶[0035]); process 400 can organize tokens into n-grams of a specified length, e.g., one, two, or three words (“one or more selectable word level n-gram lengths”) (¶[0045]: Figure 4: Step 416); implicitly, n-gram processing and stop word removal are “natural language processing”;
“identifying textual data entries of the plurality of textual data entries containing repeated n-gram instances of the identified word level n-grams” – trending topics prediction system can identify topics across billions of posts by extracting text from each post; trending topics prediction system can have each n-gram, extracted from an originating post, tagged and stored in a cumulative set of n-grams (¶[0010]);
“determining respective numbers of the identified textual data entries of the plurality of textual data entries containing repeated n-gram instances of the identified word level n-grams in the at least one data file” – frequency computer 350 can receive a cumulative set of n-grams and compute a frequency score for each unique n-gram; a frequency score within a set of a ‘unique’ n-gram is an occurrence value for all n-grams within that set that have the same sequence of tokens; in a set of n-grams ‘here we go’, ‘we’re on our way’, ‘here we go’, and ‘here we go’, where an occurrence value is a total count, there are two unique n-grams, ‘here we go’ and ‘we’re on our way’, with an occurrence value of three for ‘here we go’ and an occurrence value of one for ‘we’re on our way’; frequency computer 350 can count for each unique n-gram a number of times that unique n-gram occurs (¶[0036]: Figure 3); “determining respective numbers of the identified textual data entries” is performed because a process 400 for predicting trending topics sets a first post in Step 410 and then determines if all of the posts from the set of posts are processed by a loop in Step 418 (¶[0040] - ¶[0046]: Figure 4); that is, process 400 provides a loop that maintains a count of the set of posts that iterates to consider every post, and the set of posts contain repeated n-gram instances (“determining respective numbers of the identified textual data entries of the plurality of textual data entries containing repeated n-gram instances”);
“sorting the repeated n-gram instances based, in part, on the determined respective numbers of the identified textual data entries containing the repeated n-gram instances of the identified word level n-grams in the at least one data file occurring across the plurality of data entries to determine a most mentioned list of repeated n-gram instances for the at least one data file that is indicative of trending themes occurring across the textual data in the at least one data file” – n-grams can be sorted by their frequency score; n-grams with a frequency above a threshold, i.e., high frequency n-grams, can be passed to prediction engine 352; prediction engine 352 can receive the high frequency n-grams and compute a prediction value as an expectation of how much that n-gram will be trending in the future; n-grams can be sorted according to their computed prediction value; top scoring n-grams, e.g., n-grams with a prediction value above a threshold, can be determined to be likely trending in the future (¶[0038]: Figure 3); process 400 can compute a value for each n-gram selected representing a prediction for an amount the n-gram will be trending in the future (¶[0050]: Figure 4: Step 430); Figure 6B illustrates a predicted topic report 540 of bi-grams that includes a first column of ‘Most Discussed’ (“a most mentioned list”) (¶[0059]: Figure 6B).
Concerning independent claims 1 and 10, Tiwari discloses all of the limitations with an exception of “outputting at least the most mentioned list of repeated n-gram instances in a format adapted for being appended to the at least one data file.”
Concerning independent claims 1 and 10, Consul et al. teaches generation and use of an email frequent word list for generating a mailbox specific frequent word list and a universal frequent word list. (Abstract) A mailbox specific frequent word list 104 includes a list of frequent words found in a user’s mailbox and a corresponding frequency associated with each of the words. The list of frequent words may be sorted in order of frequency so that the most frequent words may be shown at the top of the mailbox specific frequent word list 104, and the remaining words may be shown in descending order of frequency. (Column 2, Lines 58 to 65: Figure 1) A mailbox specific frequent word list 104 can be formatted in Extensible Markup Language (‘XML’). An example of an XML data structure for an entry in the mailbox specific frequent word list 104 is <TopNWord=“____” Frequency= “_____”></TopNWord>. Other forms for representing entries in mailbox specific frequent word list 104 may be contemplated by those skilled in the art. (Column 3, Lines 4 to 18: Figure 1) API 112 generates universal frequent word list 118 by counting the number of document identifiers associated with each of the words in search data 206. If the word ‘apple’ is included in five emails, and the word ‘bear’ is included in three emails, ‘apple’ has a frequency of five and ‘bear’ has a frequency of three. Search API 112 may filter universal frequent word list 118 for only words contained in emails associated with a specific mailbox. Email server 100 maintains a mapping for each mailbox, and its corresponding emails. The mailbox specific frequent word list 104 may be formatted in XML or other suitable representation. The mailbox specific word list 104 may be stored as a folder associated item (‘FAI’) and may be represented by a data structure specifying a particular mailbox, which is identified by a mailbox identifier. An exemplary XML representation of the mailbox specific frequent word list 104 is denoted ‘TopNWords’ along with a mailbox identifier ‘mailGuid’. An exemplary XML representation of this data structure is denoted ‘WordFrequency’ and includes a ‘Word’ and its associated ‘Frequency’. (Column 5, Line 24 to Column 5, Line 37: Figure 2) Here, a mailbox specific frequent word list 104 that is stored in an XML format is “outputting at least the most mentioned list . . . in a format adapted for being appended to the at least one data file.” That is, XML or other format is “a format adapted for being appended to the at least one data file.” Broadly, Applicant’s limitation of “adapted for being appended” does not actually require that the most mentioned list is, in fact, appended, only that it is a format “adapted for being appended”. However, Consul et al. provides a mapping between a particular mailbox and a mailbox specific frequent word list 104, which is equivalent to ‘appending’ a mailbox specific frequent word list 104 to emails of the mailbox. An objective is to infer information about a user so that a user’s mailbox may be a valuable source of relevant information about the user for application programs that can utilize or benefit from this information. (Column 1, Lines 16 to 27) It would have been obvious to one having ordinary skill in the art to output a most mentioned list of repeated words in a format adapted for being appended to a data file as taught by Consul et al. to predict trending topics from n-grams of social media posts in Tiwari for a purpose of utilizing valuable information about a user from a user’s mailbox for application programs that can benefit from this information.
Concerning claims 2 and 11, Tiwari discloses that identification of n-grams likely to be trending in the future can be provided, e.g., through interface 342 (¶[0038]: Figure 3); process 400 can select a top five n-grams (¶[0051]: Figure 4); n-grams selected can be surfaced to users in a variety of ways (¶[0052]: Figure 4); Figure 6B illustrates a predicted topic report 540 (“an output analysis data file”) that includes a first column of ‘Most Discussed’ (“the results including at least the most mentioned list”) (¶[0059]: Figure 6B).
Concerning claims 7 and 16, Tiwari discloses that n-grams can be sorted by their frequency score (¶[0036]: Figure 3); n-grams are sorted according to their computed prediction value, and top scoring n-grams, e.g., n-grams with a prediction value above a threshold, can be determined as likely to be trending; identifications of these top-scoring n-grams can be provided, e.g., through interface 342 (“an ascending order list starting from a largest number of repeated n-gram instances”) (¶[0038]: Figure 3); process 400 can determine a frequency value for each unique n-gram; the frequency value can be a total count of the occurrences of the n-gram (¶[0048]: Figure 4: Step 424); process 400 can sort the n-grams based on corresponding prediction values, and select a top five n-grams (¶[0051]: Figure 4: Step 432); Figure 6B illustrates a predicted topic report 540 that includes a first column of ‘Most Discussed’ (¶[0059]: Figure 6B); broadly, a list of n-grams sorted by frequency is ‘ascending’ from the bottom to the top.
Concerning claims 8 and 17, Tiwari discloses that n-grams are sorted by their frequency score, and n-grams with a frequency score above a threshold can be passed to prediction engine 352 (“having counts above a predetermined number”) (¶[0036]: Figure 3); n-grams are sorted according to their computed prediction value, and top scoring n-grams, e.g., n-grams with a prediction value above a threshold, can be determined as likely to be trending; identifications of these top-scoring n-grams can be provided, e.g., through interface 342 (¶[0038]: Figure 3); the frequency value can be a total count of the occurrences of the n-gram within the group (¶[0048]: Figure 4: Step 424); process 400 can select a top five n-grams (¶[0051]: Figure 4).
Concerning claim 9, Tiwari discloses n-gram generator 348 can remove from a cumulative set n-grams that contain certain specified stop words; a set of stop words can be words that appear above a threshold frequency in a language, words determined to be offensive, or manually selected words, e.g., words determined to be unhelpful for determining a trending topic; stop words can include numbers (¶[0034] - ¶[0035]: Figure 3); process 400 can remove n-grams from a cumulative set of n-grams that include one or more stop words; a defined set of stop words can include any words that appear above a threshold frequency in a language, e.g., the, a, she, etc., words determined to be offensive, or manually selected words, e.g., words determined to be unhelpful for determining a trending topic; stop words can include numbers (¶[0047]: Figure 4: Step 422); here, stop words that appear above a threshold frequency in a language, e.g., the, a, she, etc., are “basic stop words”; manually selected stop words can be construed as “stop words derived from a library”, i.e., a user defines which words are to be included in the ‘library’.
Claims 3 to 5 and 12 to 14 are rejected under 35 U.S.C. 103 as being unpatentable over Tiwari (U.S. Patent Publication 2019/0102374) in view of Consul et al. (U.S. Patent No. 9,165,056) as applied to claims 1 to 2 and 11 to 12 above, and further in view of Reynolds et al. (U.S. Patent Publication 2018/0210936).
Concerning claims 3 and 12, Tiwari discloses “the output analysis file further includes information concerning a . . . column pertaining to the results of the application of NLP to the plurality of textual data entries” as illustrated in Figure 6B. Here, Tiwari discloses a predicted topic report 650 (“the output analysis data file”) has a column of ‘Most Discussed’ that is “a column pertaining to the results of the application of NLP to the plurality of textual data entries”. (¶[0059]: Figure 6B) That is, a column of ‘Most Discussed’ is what results from sorting n-gram in natural language processing. However, Tiwari does not clearly disclose that a column of predicted topic report 650 include a data “narrative” column. A ‘Most Discussed’ column in Figure 6B only appears to present a list of most frequent n-grams, but not narrative data about the n-grams.
Concerning claims 3 and 12, Reynolds et al. teaches an interactive interface for presenting insight calculations that summarize data attributes. (Abstract) Specifically, Reynolds et al. teaches that trending datasets 2130 may be disposed in a portion of a user interface 2102 that presents user interface elements including text information 2152 describing trending dataset 2150. A description may include a purpose of a dataset, a source of a dataset, and a field of applicability. (¶[0170]: Figures 20 to 21) Here, Figure 20 includes a column of recent discussion 2024 with “narrative” descriptions of a trending dataset, e.g., ‘Hate Crime Laws and Statistics’, ‘TED Talks Complete List’, etc. An objective is to present summarization of dataset attributes to facilitate discovery, formation, and analysis of interrelated collaborative datasets. (¶[0002]) It would have been obvious to one having ordinary skill in the art to provide a narrative description as taught by Reynolds et al. of a column pertaining to results of natural language processing in Tiwari for a purpose of facilitating discovery, formation, and analysis of interrelated collaborative datasets.
Concerning claims 4 and 13, Tiwari discloses “a plurality of textual entries”, but omits “a separate cell within the at least one data file”. Here, Tiwari is directed to receiving textual entries as posts, but discloses a variety of sources of textual information. (¶[0013]) Generally, it is known that electronic spreadsheets include textual entries in ‘cells’, so it would be merely a matter of an application to electronic spreadsheets in Tiwari. Specifically, Reynolds et al. teaches a data file 601a may be received in a variety of formats, and dataset analyzer 630 may be configured to analyze data file 601a to detect and resolve if a cell contains useful data including a string in a column of a tabular data file. (¶[0090] - ¶[0091]) Reynolds et al., then, teaches that tabular data could be presented within cells. An objective is to present summarization of dataset attributes to facilitate discovery, formation, and analysis of interrelated collaborative datasets. (¶[0002]) It would have been obvious to one having ordinary skill in the art to perform natural language processing on n-grams of Tiwari for tabular data in cells as taught by Reynolds et al. for a purpose of facilitating discovery, formation, and analysis of interrelated collaborative datasets.
Concerning claims 5 and 14, Reynolds et al. teaches that a data file 601a may be received in a variety of formats, and dataset analyzer 630 may be configured to analyze data file 601a to detect and resolve if a cell contains useful data including a string in a column of a tabular data file (“wherein each of the separate cells are part of a same data narrative column within the at least one data file”). (¶[0090] - ¶[0091])
Claims 6 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Tiwari (U.S. Patent Publication 2019/0102374) in view of Consul et al. (U.S. Patent No. 9,165,056) as applied to claims 1 to 2 and 11 to 12 above, and further in view of Udupa et al. (U.S. Patent Publication 2013/0151533).
Tiwari discloses “sorting the repeated n-gram instances based, in part, on the determined numbers of the identified textual data entries containing the repeated n-gram instances of the identified word level n-grams in the at least one data file occurring across the plurality of textual data entries”, but omits “searching for nouns and adjectives near each identified n-gram in a textual data entry of the plurality of textual data entries pertaining to the identified n-gram.” However, Udupa et al. teaches extracting n-grams from a document corpus, where a size of n-grams extracted from each document can range between one and five. A well-formedness score can be computed for each n-gram from documents, wherein the well-formedness score is indicative of parts of speech in the analyzed n-gram as well as the arrangement of parts of speech in the n-gram. One or more natural language processing algorithms can be employed in connection with computing the well-formedness score, or a series of rules can be analyzed with respect to an n-gram to compute the well-formedness score. Specifically, an n-gram that begins with a noun, a verb, or a participle and ends with a noun or a participle will be provided with a relatively high well-formedness score, while an n-gram that begins or ends with an adjective or adverb will be provided with a relatively low well-formedness score. Each n-gram that has a well-formedness score above a predefined threshold can be retained as a candidate key phrase in a list of candidate key phrases. An objective is to identify key phrases in documents. (¶[0007] - ¶[0008]) It would have been obvious to one having ordinary skill in the art to analyze n-grams to determine nouns and adjectives near each identified n-gram in a textual data entry as taught by Udupa et al. to predict future trending topics from n-grams in Tiwari for a purpose of identifying key phrases in documents.
Response to Arguments
Applicant's arguments filed 18 May 2026 have been fully considered but they are not persuasive.
Applicant does not provide any amendments to the claims, but presents arguments traversing the prior rejection of the independent claims as being obvious under 35 U.S.C. §103 over Tiwari (U.S. Patent Publication 2019/0102374) in view of Consul et al. (U.S. Patent No. 9,165,056).
Generally, Applicant argues that Tiwari is ‘fundamentally different’ from the invention because it is designed to measure “gross popularity” of an n-gram across an entire corpus, but the invention identifies an “intra-entry significance” of an n-gram before next performing a “cross-entry analysis” measuring its prevalence. Specifically, Applicant states that Tiwari aggregates all n-grams from all posts into a single collection and then performs a simple frequency count. Applicant alleges that Tiwari has no provision for checking for repeats within the original posts during the collection phase. Applicant argues that the rejection is predicated on a “subjective inference” that trending topics of Tiwari are equivalent to Applicant’s ‘repeated n-grams’, and that this is an “unsupported inference baldly asserting” this equivalence, which Applicant maintains is not a tenable argument. Applicant states that the rejection misinterprets Tiwari, where each post is processed individually, and that the rejection improperly conflates the counting of total occurrences of n-grams with the counting of the number of entries that contain repeated instances of an n-gram. Consequently, Applicant argues that these are fundamentally different metrics with a “gross frequency count” being explicitly disclosed by Tiwari. Specifically, Applicant sets forth three cases designated (1) Case A where one post contains ‘here we go’ three times, (2) Case B where one post contains ‘here we go’ twice and a second post contains it once, and (3) Case C where three different posts each contain ‘here we go’ one time. Applicant argues that the invention can be contrasted with Tiwari which would ignore n-grams in Case C because there are no repeated n-grams. Applicant characterizes Tiwari as providing an architecture which “dumps all n-grams into one large undifferentiated ‘cumulative’ set”. Applicant admits that both systems count n-grams, but maintains how they are counting is distinct and nonobvious. Applicant characterizes Tiwari as counting total occurrences from a “flattened list”, but the invention counts entries that have passed a preliminary test of “intra-entry repetition”, i.e., repeated n-gram in an entry. Applicant alleges that Tiwari has a goal of finding what is popular, but the invention finds what is problematic or thematic by identifying a phrase that an author deemed important enough to repeat within a single descriptive entry.
Then Applicant provides arguments directed against Consul et al. Applicant argues that the context of Consul et al. is entirely different from Tiwari. Here, Applicant argues that Consul et al. is directed to analyzing a private email mailbox to infer an individual user’s interest for purposes of targeted advertising, so that there is a one-to-one profiling, and not a many-to-one system for analyzing trending topics. Additionally, Applicant contends that there would be no motivation to combine Consul et al. with Tiwari because the data sources are different, public versus private, and the scales are different, thousands or millions of users versus one user.
Applicant’s arguments are being carefully considered, but are unpersuasive, and the rejection of the independent claims in being maintained for obviousness under 35 U.S.C. §103 over Tiwari (U.S. Patent Publication 2019/0102374) in view of Consul et al. (U.S. Patent No. 9,165,056). The main problem with Applicant’s arguments is that they are directed to features that are unclaimed. Applicant throws around a variety of terms, e.g., “gross popularity”, “intra-entry significance”, and “cross-entry analysis”, but these terms are not set forth by the claim language and are not supported by the Specification. Applicant attempts to create a sophisticated analysis comparing the prior art to the claim language, but the arguments fails to consider what is taught by Consul et al. Although the claims are interpreted in light of the specification, limitations from the specification are not read into the claims. See In re Van Geuns, 988 F.2d 1181, 26 USPQ2d 1057 (Fed. Cir. 1993).
The independent claims, in fact, only set forth two or three broad terms upon which Applicant attempts to interpret the scope of the claim language to distinguish over the prior art – “a plurality of textual data entries”, “word-level n-grams”, and “repeated n-gram instances”. The claim language as broadly construed cannot distinguish over the prior art merely by the use of these three terms. Specifically, Applicant’s claim language does not distinguish a counting mechanism that operates in a ‘fundamentally different way’ as characterized by Applicant as compared to Tiwari and Consul et al. That is, even if one were to accept the characterization of the counting mechanism of the three cases, Case A, Case B, and Case C as created by Applicant, the claim language does not distinguish over these distinctions. Here, it is axiomatic that an n-gram is a repeated n-gram if the string of the n-gram is identical to another n-gram. Consequently, Tiwari is counting n-grams that are the same, as given by the string ‘here we go’. The n-gram ‘here we go’ is a repeated n-gram precisely because it is the same as another occurrence of an n-gram ‘here we go’, and an n-gram ‘we are on our way’ is not repeated and not counted as the same n-gram. Applicant’s claim language does not specify whether or not an n-gram must be repeated in the same “textual entry” in order to count or whether or not an n-gram can be repeated in a different “textual entry” in order to count. This is the crux of Applicant’s argument with Case A, Case B, and Case C. However, Applicant’s claim language, as broadly construed does not provide any distinction as to how the n-gram must be counted. Applicant, then, is attempting to construct a sophisticated argument of a distinction between the invention and Tiwari, but there is no basis for the distinction under a broad construction of the claim language.
During patent examination, the pending claims must be “given their broadest reasonable interpretation consistent with the specification.” Phillips v. AWH Corp., 415 F.3d 1303, 1316, 75 USPQ2d 1321, 1329 (Fed. Cir. 2005) Because applicant has the opportunity to amend the claims during prosecution, giving a claim its broadest reasonable interpretation will reduce the possibility that the claim, once issued, will be interpreted more broadly than is justified. In re Yamamoto, 740 F.2d 1569, 1571 (Fed. Cir. 1984); In re Zletz, 893 F.2d 319, 321, 13 USPQ2d 1320, 1322 (Fed. Cir. 1989) (“During patent examination the pending claims must be interpreted as broadly as their terms reasonably allow.”); In re Prater, 415 F.2d 1393, 1404-05, 162 USPQ 541, 550-51 (CCPA 1969) See MPEP §2111.
Applicant’s claim language sets forth “at least one data file containing a plurality of textual entries”. By this claim language, there could be one page of text (“one data file”) containing a plurality of individual strings of text (“a plurality of textual data entries”). Or there could be thousands of documents in “at least one data file” and each document could be one “data entry”. These terms can be broadly construed. Now the claim language sets forth “word level n-grams” and “repeated n-gram instances of these word level n-grams”. Here, “word level n-grams” are just a sequence of consecutive words in “textual data entries”. The “word level n-grams” could all be on the same page of text in a plurality of sentences of the text with strings of words in the sentences serving as “a plurality of textual data entries”. Or, Applicant’s “word level n-grams” could be a sequence of consecutive words that occur over thousands of documents with each of the documents serving as a textual entry in the thousands of documents that are “the plurality of textual entries”. There is no way to distinguish these interpretations given only the claim language.
Let’s consider Applicant’s point that the n-grams of the invention are counted in a different way. What does the claim language read is being counted? If it is the “textual data entries” containing the repeated n-grams that are being counted, does this necessarily produce a different result in a number being counted than if the n-grams are being counted? This depends upon how one construes “textual data entries”. If the “textual data entries” are just construed as strings of words in sentences or some set of the words in a document, does it still make a difference? Is this distinction clearly set forth by the claim language? The claim language states:
“determining respective numbers of the identified textual data entries of the plurality of textual data entries containing the repeated n-gram instances of the identified word level n-grams in the at least one data file;
sorting the repeated n-grams instances based, in part, on the determined respective numbers of the identified textual data entries . . . .”
Reading these limitations broadly, does the claim language clearly state that what is being counted are “identified textual data entries” or “repeated n-gram instances” that happen to be in “identified textual data entries”? If one does not construe “identified textual data entries” as being different documents, does it make a difference? Couldn’t one just construe “identified textual data entries” as strings of words that may or may not contain “repeated n-gram instances”? That is, some “identified textual data entries” could just be strings of words that do not happen to contain “repeated n-gram instances” in which case they are not counted, and some “identified textual data entries” could be strings of words that do contain “repeated n-gram instances” in which case they are counted. Broadly, “identified textual entries” does not necessarily have to be construed as different documents but could be different strings of words in a given document.
Nevertheless, we can now look to what is taught by Consul et al. Specifically, Consul et al., at Column 5, Lines 29 to 36:
In one embodiment, the API 112 generates the universal frequent word list 118 by counting the number of document identifiers associated with each of the words in the search data 206. For example, in the example shown above, the word “apple” is included in five emails, while the word “bear” is included in three emails. As such, “apple” has a frequency of five, and “bear” has a frequency of three.
Here, Consul et al. is teaching a counting mechanism that is exactly the same as that of the invention as characterized by Applicant. Consul et al. does not count the total number of occurrences of the term “bear” or the term “apple”, but instead counts the number of emails with the word “bear” and the number of emails with the word “apple”. The number of e-mails with the word “bear” is three and the number of emails with the word “apple” is five. So, even if one of the emails had the word “bear” fifty times, the count of a repeated word “bear” would still only be three because that is the number of emails that have at least one occurrence of the word “bear”. Similarly, Consul et al. states, at Column 2, Line 61 to Column 3, Line 3:
The list of frequent words may be sorted in order of frequency. For example, the most frequent words may be shown at the top of the mailbox specific frequent word list 104, and the remaining words may be shown in a descending order of frequency. The frequency may be specified as a raw frequency (e.g., the absolute number of email messages that include a word) or a percentage/ratio (e.g., the number of email messages that include a word in relation to the total number of messages across the user's mailbox). (emphasis added)
Again, a count is based on a number of emails messages that include a repeated word, and not necessarily on a number of times that a word is repeated across all email messages in Consul et al. At Column 7, Lines 17 to 21, Consul et al. states:
In one embodiment, the universal frequent word list 118 includes a mapping of the words to a frequency associated with each of the words across multiple mailboxes. The frequency may be determined by counting the number of email identifiers corresponding to each of the words. (emphasis added)
This is the point argued by Applicant, so that even if one were to construe the claim language in a manner that is advocated by Applicant, Consul et al. still provides a teaching for modification to address this argument. Consul et al. does not just count a number of times a word is repeated, but counts the number of email identifiers that contain the word that is repeated.
Applicant’s argument is not persuasive that there is no motivation to combine Consul et al. with Tiwari. Applicant is only arguing the specific teachings of the two references individually without consideration for what the combination suggests to one having ordinary skill in the art. One cannot show nonobviousness by attacking references individually where the rejections are based on combinations of references. See In re Keller, 642 F.2d 413, 208 USPQ 871 (CCPA 1981); In re Merck & Co., 800 F.2d 1091, 231 USPQ 375 (Fed. Cir. 1986).
Here, it is not as if the two reference are from different fields of endeavor. Both Tiwari and Consul et al. are directed to counting repeated instances of words to determine the most frequent words that can then be used for practical purposes. Specifically, Tiwari, at ¶[0001] - ¶[0002] and ¶[0034], states that trending topics can be used for advertisers to select content for advertisers to display. Identically, Consul et al., at Column 3, Lines 19 to 43, teaches an objective of an advertising application that tailors advertisements to a user based on frequent words in a mailbox. So, Applicant’s argument is not persuasive that a context is entirely different for Tiwari and Consul et al. There may always be individual differences between the specifics of two references in a combination but that does not prevent a reasonable combination of these references. Here, objectives of the two references are the same, i.e., for advertising purposes, and it is not necessary that this objective must be same as that of the invention.
The rejection sets forth an express motivation for a combination in Consul et al.: To infer information about a user so that a user's mailbox may be a valuable source of relevant information about the user for application programs that can utilize or benefit from this information. See Column 1, Lines 16 to 27 of Consul et al. However, a rejection can similarly be predicated on a rationale under SR International Co. v. Teleflex Inc. (KSR), 550 U.S. 398, 82 USPQ2d 1385 (2007): (A) Combining prior art elements according to known methods to yield predictable results. Consul et al. provides a known method of counting a number of emails having frequent words. It would be predictable result to one skilled in the art to provide a variation in a manner in which frequent words are counted in Tiwari according to way frequent words are counted in Consul et al.
Applicant’s arguments are not persuasive. There are no new grounds of rejection. This Office Action is properly FINAL.
Conclusion
The prior art made of record and not relied upon is considered pertinent to Applicant’s disclosure.
Chang et al. is additional prior art that determines trending topics from text communications after stopword removal and stemming for N-grams of various lengths by term frequency (TF) ranking. See Figure 4.
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MARTIN LERNER whose telephone number is (571) 272-7608. The examiner can normally be reached Monday-Thursday 8:30 AM-6:00 PM.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Richemond Dorvil can be reached on (571) 272-7602. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/MARTIN LERNER/Primary Examiner
Art Unit 2658 June 23, 2026
Prosecution Insights