Detailed Action
Notice of Pre-AIA or AIA status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Arguments
The objection to claim 1, and rejection of claims 8 and 18 under 35 U.S.C. § 112(b), are each withdrawn, responsive to the Applicant’s amendments correcting the problems identified in the objection and the rejection.
The present response does not attempt to amend the claims to resolve the issues raised under 35 U.S.C. §§ 101, 103, or 112(a), and none of the Applicant’s arguments concerning those rejections persuade the Examiner of error. Accordingly, since the rejections under 35 U.S.C. §§ 101, 103, and 112(a) are each correct, they stand. Each of the Applicant’s arguments will now be addressed.
New Matter
Claims 7–9 and 16–18 stand rejected under 35 U.S.C. § 112(a) or 35 U.S.C. § 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The Applicant’s traversal of the rejection has been considered, but is not persuasive.
While the Applicant correctly points out that the specification discloses the concepts of multiple thresholds and pre-defined clusters in Figures 6A and 6C, the arguments fail to overcome the rejection because they ignore the rejection’s main finding that the scope of claims 7–9 requires a system that generates two different pluralities of clusters of thought objects for the same input plurality of text segments. The specification doesn’t support this.
For example, consider what happens when the system of claim 7 receives a plurality of thought objects above the threshold—keeping in mind that the entire text of claim 1 is incorporated into claim 7: First, as required by the text of claim 1, we compute a reduced plurality of the received thought objects. Next, we cluster the reduced plurality of thought objects into a number of clusters that is based on length of the desired summary and the quantity of reduced thought objects. Then, we generate and display a summary using clusters we just discussed.
After having already reduced and clustered the thought objects, and generated the summary using the clusters, we pick up where we left off in claim 7, we look back at the original set of thought objects (rather than the reduced set), notice that their quantity exceeds the threshold, and then create a new, pre-defined number of clusters, and use this second set of clusters to generate and transmit a headline for the originally-received plurality of thought objects.
The specification describes these clustering operations as mutually exclusive operational pathways, such as generating a summary versus generating a headline. The specification nowhere describes a single embodiment where the system executes both operational pathways simultaneously on the exact same plurality of thought objects. Because the literal language of the claims requires an impossible, hybrid workflow that contradicts the specification’s flowcharts, the claims recite an invention that lacks written description support.
Accordingly, the rejection of claims 7–9 and 17–19 under 35 U.S.C. § 112(a) is maintained.
Subject-Matter Eligibility
Claims 1–20 stand rejected under 35 U.S.C. § 101 because the claimed invention remains directed to a judicial exception without significantly more.
The Applicant argues that the claims represent a specific technological improvement in computer-implemented summarization systems, asserting that determining the number of semantic clusters based on summary length and input objects, and clustering using semantic vector representations, cannot practically be performed by the human mind. (Response 10). The Examiner respectfully disagrees.
Merely claiming an invention that uses a generic computer to perform calculations faster than a human could perform them is not the same as an invention that actually improves the functioning of a computer such that the computer performs calculations faster or more efficiently than a pre-solution computer. CLS Bank Intern. v. Alice Corp. Pty. Ltd., 717 F. 3d 1269, 1286 (Fed. Cir. 2013) (Lourie, J., concurring), aff’d Alice Corp. v. CLS Bank International, 573 U.S. 208, 223–24 (2014). The former merely relies on the inherent speed and processing power of conventional computer hardware to automate abstract ideas, such as mathematical formulas or mental processes. The latter involves a specific, technical improvement to the computer system itself, such as a novel data structure or a fundamentally new way of allocating memory, that enhances the machine's core operational capabilities.
The present claims fall squarely into the former category. The core activities recited in the claims—filtering redundant information, organizing text into meaningful groups, and selecting representative text for a summary—are fundamental cognitive and mental processes. Representing text as semantic vectors is simply the application of a mathematical concept to organize and manipulate this textual data. The claims do not recite any structural changes to how the computer operates or processes data at a fundamental level. Rather, they merely employ a generic processor and memory to execute mathematical clustering algorithms on textual data more quickly and efficiently than a human reader could. Merely automating these human-performable tasks using mathematical constructs to achieve faster processing times does not remove the claims from the realm of judicial exceptions.
The Applicant also relies on Enfish LLC v. Microsoft Corp. and McRO, Inc. v. Bandai Namco Games Am. Inc. to argue that the claims provide an unconventional architecture that improves computer performance and integrates the abstract idea into a practical application. (Response 11). This reliance is unpersuasive. In Enfish, the claims were eligible because they recited a specific, structural improvement to how a computer physically stores and retrieves data in memory via a self-referential logical table. The present claims, however, merely invoke generic computer components to carry out the abstract summarization process. The alleged improvement here relates to the mathematical algorithm itself, rather than any efficiencies that a computer gains from executing them.
Furthermore, the present claims are distinguishable from the claims in McRO. In McRO, the claims were held eligible because they recited a highly specific, detailed set of mathematical rules that replaced a subjective human process with an automated, objective methodology. The instant claims do not recite a specific technological rule set. Instead, the claims recite high-level, result-oriented functional steps, such as computing a reduced plurality of thought objects and generating a plurality of clusters by analyzing a semantic vector representation. The claims describe what the system achieves—generating a quantity of clusters based on length and quantity—rather than the specific, technical algorithmic rules for how it achieves that generation. Because the claims are directed to an abstract idea and merely recite generic computer components functioning in their routine and conventional capacities to execute mathematical calculations and mental steps, they fail to integrate the abstract idea into a practical application and do not provide significantly more than the abstract idea itself.
Therefore, the rejection of claims 1–20 under 35 U.S.C. § 101 is maintained.
Obviousness
Claims 1, 4–6, 10–11, 16–16, and 20 stand rejected under 35 U.S.C. § 103 as being unpatentable over U.S. Patent Application Publication No. 2022/0215052 A1 (“Chalana”) in view of De-Xi Liu et al., A Novel Chinese Multi-Document Summarization Using Clustering Based Sentence Extraction, Proceedings of the Fifth International Conference on Machine Learning and Cybernetics (August 2006) (“Liu”).
The Applicant contends that combining Liu’s heuristic with Chalana’s system would require dismantling Chalana’s attention-based clustering layer and replacing it with an explicit K-means function, which, in the Applicant’s opinion, would destroy Chalana’s feed-forward neural architecture and change its principle of operation. (Response 20).
This argument mischaracterizes the nature of the proposed combination. The test for obviousness does not require the bodily incorporation of the components of one reference into another. Instead, the test is whether the teachings of the references, as a whole, would suggest the claimed invention to a person of ordinary skill in the art. In re Nievelt, 482 F.2d 965, 968 (CCPA 1973); In re Keller, 642 F.2d 413, 425 (CCPA 1981).
Chalana explicitly teaches that its neural network determines a number of sentence meaning clusters “based on a desired length of a summary.” Chalana ¶ 84. The proposed combination simply modifies how that target number of clusters is calculated prior to executing Chalana’s neural network. Implementing Liu’s mathematical heuristic,
K
=
L
S
M
/
a
v
g
L
S
, to calculate this variable does not require replacing Chalana’s neural network with a K-means algorithm. The transformer architecture remains entirely intact; it simply receives a more accurately calculated cluster target based on the specific document’s average sentence length and total sentence count, exactly as taught by Liu. Because the fundamental operation of Chalana’s neural network is preserved, the principle of operation is not destroyed.
The Applicant also argues that Chalana does not disclose any step that first reduces the input corpus before clustering. (See Response 20) (“Nor does Chalana disclose any step that first reduces the input corpus before clustering; its clustering arises internally through transformer attention mechanisms and not through an external, parameterized K-means process.”).
Respectfully, this is simply untrue. Chalana explicitly teaches receiving user feedback comprising “an instruction to include or exclude a portion of the audio-visual media from at least one of the transcript [or] the transcript summary.” Chalana ¶ 30. Furthermore, Chalana explicitly teaches implementing these changes by excluding the selected portions of the transcript and inputting the “modified version transcript” into the transcription summarization module 600. Chalana ¶¶ 65 and 66. This unequivocally teaches computing a reduced plurality of thought objects prior to executing the clustering module.
The Applicant also argues that “Liu’s static K-means routine is inapplicable to Chalana’s dynamic transformer environment without undue experimentation.” (Response 21). The Examiner respectfully disagrees.
As explained above, the combination does not require integrating a K-means algorithm into a neural network. The combination simply requires passing a dynamically calculated integer (the number of clusters) to a neural network that is already designed to receive and act upon a target cluster count. A person of ordinary skill in the art of machine learning and natural language processing would possess the routine programming skills necessary to calculate a variable using basic division and provide that variable to a neural network parameter, requiring no undue experimentation.
Lastly, the Applicant’s argument that the Examiner’s rationale is conclusory and lacks the articulated reasoning required by In re Kahn is not persuasive, because it’s not true.
The Office Action provided clear, articulated reasoning anchored directly in the express teachings of the prior art. (See Non-Final Office Action 15 ¶ 64) (quoting directly from Liu 2594). The Examiner pointed out that Chalana generates clusters based merely on desired length. The Examiner then pointed to Liu, which expressly teaches that the “most probable number of sentences in a fixed-length-summary is the length of summary divided by the average length of sentences in document collection.” (Liu, page 2594).
The rational underpinning is straightforward: one of ordinary skill in the art would be motivated to improve Chalana’s cluster-generation parameter using Liu’s specific heuristic because calculating the average sentence length provides a mathematically superior target for the number of clusters required to fill a fixed-length summary. Applying this known heuristic to Chalana mathematically optimizes the cluster generation process to prevent generating too many or too few clusters for the requested length. This constitutes a predictable use of prior art elements according to their established functions.
Accordingly, the rejection of claims 1, 4–6, 10–11, 14–16, and 20 under 35 U.S.C. § 103 is maintained. No additional arguments were raised for the other 35 U.S.C. § 103 rejections, so they stand as well.
Since all of the claims stand rejected, the request for a notice of allowance (Response 22) is respectfully denied.
Claim Rejections – 35 U.S.C. § 112(a)
The following is a quotation of 35 U.S.C. § 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of 35 U.S.C. § 112 (pre-AIA ), first paragraph:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claims 7–9 and 16–18 are rejected under 35 U.S.C. § 112(a) or 35 U.S.C. § 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for pre-AIA the inventor(s), at the time the application was filed, had possession of the claimed invention.
Claim 7
The preliminary amendment to claim 7, taken together with the preliminary amendment to claim 1 (which claim 7 incorporates by reference) recites an invention that was not disclosed in the original Written Description.
Claim 7, as currently written, requires the system to generate two different sets of clusters of thought objects for the same plurality of text segments. This is because claim 7 refers back to the original plurality of thought objects, rather than the reduced plurality of thought objects. So, reading claim 7 together with all of the limitations it incorporates from claim 1 by reference, the claimed system (1) receives an original plurality of thought objects; (2) generates a pre-defined number of clusters of the original thought objects; (3) computes a reduced plurality of thought objects; and (4) generates a plurality of clusters of thought objects from the reduced plurality of thought objects.
In contrast, FIG. 6A and the associated text shows that the system only generates (2) or (4), not both, depending on the number of original thought objects and the requested length.
Claim 8
Claim 8 recites new matter at least because it incorporates the new matter of its parent claim by reference.
Claim 9
The preliminary amendment to claim 9, taken together with the preliminary amendment to claim 1 (which claim 9 incorporates by reference) causes the following element of claim 9 to recite new matter:
responsive to determining that the plurality of thought objects is above a first threshold and below a second threshold, generate a pre-defined number of clusters;
The above claim element contains two different pieces of new matter. For one, the Written Description never discloses using the “pre-defined number of clusters” approach responsive to determining that the plurality of thought objects are below any threshold, let alone responsive to determining that they are between two thresholds.
To the contrary, as shown in FIG. 6C and explained in paragraph 106, the specification says to compute a pre-configured number of clusters responsive to the thought objects not being less than the second threshold (i.e., 625 is performed responsive to 621 resolving as “No”). The only other instance of using a pre-configured number of clusters is in step 616 (FIG. 6A), but the determining that the plurality of thought objects are “below a second threshold” is not one of the conditions precedent to step 616. The only conditions precedent to 616 are that the thought objects are greater than the first threshold (612: Yes), and that the requested summary length is not greater than a summary length threshold (614: No). See Spec. ¶¶ 95–97.
The second piece of new matter is that claim 9, as currently written, requires the system to generate two different sets of clusters for the same plurality of text segments. This is because claim 9 refers back to the original plurality of thought objects, rather than the reduced plurality of thought objects. So, reading claim 9 together with all of the limitations it incorporates from claim 1 by reference, the claimed system (1) receives a plurality of thought objects; (2) generates a pre-defined number of clusters of the received thought objects; (3) computes a reduced plurality of thought objects; and (4) generates a plurality of clusters of thought objects from the reduced plurality of thought objects.
Claims 17–19
Claims 17–19 are rejected for the same reasons as given above for corresponding claims 7–9.
Claim Rejections – 35 U.S.C. § 101
35 U.S.C. § 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1–20 are rejected under 35 U.S.C. § 101 because the claimed invention is directed to a judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea) without significantly more.
The Office applies a four-part test when examining claims for subject-matter eligibility under § 101. First, the claimed invention must be directed to one of the four statutory categories explicitly listed in § 101 (Step 1). MPEP § 2106(I.). Then, the claimed invention is analyzed to determine whether it is directed to one of § 101’s judicial exceptions (Step 2A Prong One) without reciting both a practical application of the judicial exception (Step 2A Prong Two) and significantly more than the judicial exception (Step 2B). MPEP § 2106(I.).
With this framework in mind, the claims will now be analyzed for subject matter eligibility under § 101.
Claim 1
Step 1. Claim 1 is directed to a "system" comprising a processor and memory, which is a statutory "machine."
Step 2A, Prong One. However, the claimed invention is directed to an abstract idea, which is a judicial exception to 35 U.S.C. § 101. Specifically, the claim recites the steps of analyzing semantic vector representations to generate clusters and generating a summary therefrom. This process of analyzing data and creating a summary represents a combination of two abstract ideas falling into two enumerated groupings: mental processes and mathematical concepts.
Summarizing a body of text by identifying its core ideas and re-stating them in a condensed form is a fundamental cognitive task that a human can perform. The claimed steps of analyzing semantic meaning to form groups (clusters) and then selecting representative items from those groups to form a summary are analogous to the mental steps a person would take when reading an article, identifying the main themes, and writing an abstract. While the claims recite using a processor, the process itself, under its broadest reasonable interpretation, covers performance of a task that can be practically performed in the human mind.
Furthermore, the process recites mathematical concepts. The claims require "analyzing a semantic vector representation" and "generating a plurality of clusters" based on that analysis. A semantic vector representation is a mathematical construct, and analyzing it to form clusters involves mathematical calculations and algorithms. This is a mathematical concept used to manipulate data, which is an abstract idea.
Step 2A, Prong Two. This judicial exception is not integrated into a practical application. The claims recite additional elements, including a "text-summary system" with a processor and memory and displaying the summary on a graphical user interface. However, these elements do not integrate the abstract idea of summarization into a practical application.
The system components (processor, memory, graphical user interface) are recited at a high level of generality and serve only as a generic environment for performing the abstract summarization task. Such generic computer components amount to no more than mere instructions to apply the exception using a generic computer. These additional elements, whether taken individually or as an ordered combination, do not impose any meaningful limits on practicing the abstract idea. Accordingly, the claim is directed to an abstract idea.
Step 2B. The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception. As discussed above with respect to the integration into a practical application, the additional elements recited in the claims are generic computer components. The recitation of a generic processor and memory to perform the analysis and summarization amounts to no more than mere instructions to apply the exception on a computer. The data gathering and display steps are insignificant extra-solution activity. These elements do not provide an inventive concept. The claim is not patent eligible.
Claim 2
Claim 2 is rejected under 35 U.S.C. § 101. The claim depends from claim 1 and is directed to the same abstract idea of summarization. The additional limitation of identifying and removing redundant thought objects does not provide any additional elements to analyze under Step 2A Prong Two or Step 2B, because this step is an abstract data filtering and manipulation task—it is merely another step of the mental process described in the rejection of claim 1. Adding this further abstract step does not integrate the core abstract idea of claim 1 into a practical application or add significantly more to the claim as a whole.
Claim 3
Claim 3 is rejected under 35 U.S.C. § 101. The claim depends from claim 1 and is directed to the same abstract idea of summarization. The additional limitation of generating a random sample from the thought objects when their number exceeds a threshold does not provide any additional elements to analyze under Step 2A Prong Two or Step 2B, because random sampling is a mathematical concept, and the decision to employ random sampling after counting that the number of thought objects exceeds a threshold is a mental step.
Claim 4
Claim 4 is rejected under 35 U.S.C. § 101. The claim depends from claim 1 and is directed to the same abstract idea of summarization. The additional limitation of associating a cluster to a portion of the reduced plurality of thought objects does not provide any additional elements to analyze under Step 2A Prong Two or Step 2B, because associating elements with their clusters is inherent to the definition of clustering; it is the fundamental action of any clustering process. It does not add anything beyond the abstract idea of “generating a plurality of clusters” already recited in claim 1.
Claim 5
Claim 5 is rejected under 35 U.S.C. § 101. The claim depends from claim 1 and is directed to the same abstract idea of summarization. The additional limitation of selecting thought objects using confidence scores, theming, sentiment analysis, or rating processes does not provide any additional elements to analyze under Step 2A Prong Two or Step 2B, because these selection criteria are themselves abstract ideas—either mathematical calculations (scores, ratings) or mental processes (theming, sentiment analysis). Adding more abstract ideas to the main abstract idea does not make the claim patent-eligible.
Claim 6
Claim 6 is rejected under 35 U.S.C. § 101. The claim depends from claim 5 and is directed to the same abstract idea. The additional limitation of defining confidence scores as indicative of a "quantified importance" does not provide any additional elements to analyze under Step 2A Prong Two or Step 2B, because this limitation merely provides a mathematically formalized definition for the abstract "confidence scores" recited in claim 5, reinforcing their abstract nature. It adds no inventive step or concrete application.
Claim 7
Claim 7 is rejected under 35 U.S.C. § 101. The claim depends from claim 1 and is directed to the same abstract idea of summarization. The additional limitation of generating a "headline" based on a threshold does not provide any additional elements to analyze under Step 2A Prong Two or Step 2B, because generating a headline is the same as generating a summary whose length is a single sentence. The steps of generating clusters, selecting, and transforming objects into a headline are still part of the abstract summarization process. The conditional logic ("responsive to determining . . . a threshold") is an abstract rule that does not add significantly more.
Claim 8
Claim 8 is rejected under 35 U.S.C. § 101. The claim depends from claim 7 and is directed to the same abstract idea. The additional limitation of transforming the thought objects into a headline when their plurality is below a second threshold does not provide any additional elements to analyze under Step 2A Prong Two or Step 2B, because it simply adds another abstract logical rule governing the abstract process recited in claim 7.
Claim 9
Claim 9 is rejected under 35 U.S.C. § 101. The claim depends from claim 1 and is directed to the same abstract idea of summarization. The additional limitation of generating a summary based on conditional logic (if the number of objects is between two thresholds) does not provide any additional elements to analyze under Step 2A Prong Two or Step 2B. Instead, claim 9 adds another set of abstract rules to apply as part of the mental process.
Claim 10
Claim 10 is rejected under 35 U.S.C. § 101. The claim depends from claim 1 and is directed to the same abstract idea of summarization. The additional limitation of generating a "pre-configured number of clusters" does not provide any additional elements to analyze under Step 2A Prong Two or Step 2B; it simply defines some of the variables under consideration of the recited mental and mathematical processes.
Claims 11–20
Claims 11–20 recite the same methods that the systems of respective claims 1–10 perform as part of their normal operation. Accordingly, claims 11–20 are rejected for the same corresponding reasons, but with Step 1 of the analysis acknowledging the claims are directed to a “process” under 35 U.S.C. § 101, rather than a machine.
Claim Rejections – 35 U.S.C. § 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 of this title, 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.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned at the time any inventions covered therein were effectively filed absent any evidence to the contrary. Applicant is advised of the obligation under 37 C.F.R. § 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned at the time a later invention was effectively filed in order for the examiner to consider the applicability of 35 U.S.C. § 102(b)(2)(C) for any potential 35 U.S.C. § 102(a)(2) prior art against the later invention.
I. Chalana and Liu teach claims 1, 4–6, 10–11, 16–16, and 20.
Claims 1, 4–6, 10–11, 16–16, and 20 are rejected under 35 U.S.C. § 103 as being unpatentable over U.S. Patent Application Publication No. 2022/0215052 A1 (“Chalana”) in view of De-Xi Liu et al., A Novel Chinese Multi-Document Summarization Using Clustering Based Sentence Extraction, Proceedings of the Fifth International Conference on Machine Learning and Cybernetics (August 2006) (“Liu”).
Claim 1
Chalana teaches
A system for summarizing a plurality of text segments, the system comprising: a text-summarytext-summary system comprising at least one processor, at least one memory, and a plurality of programming instructions, the plurality of programming instructions when executed by the at least one processor causes the at least one processor to:
As shown in FIG. 1, system 100 includes an “audio-visual media synopsis computer device apparatus 200” with several program modules 400–800 whose functionality will be discussed herein. Chalana ¶ 18. As shown in FIG. 2, the apparatus 200 further comprises at least one processor (e.g., chipset 255 and/or cpu 215) for executing the program modules.
receive a plurality of thought objects, the plurality of thought objects each comprising a [text segment] from the plurality of text segments;
“At block 500, audio-visual media synopsis module 400 may call or trigger execution of, for example, audio-visual media transcription module 500. Audio-visual media transcription module 500 may, for example, prepare a transcript of video 305,” Chalana ¶ 60, which “may be saved as one or more transcript 335 records.” Chalana ¶ 61.
receive a length of a summary;
“At block 410, audio-visual media synopsis module 400 may obtain a desired length of a video summary.” Chalana ¶ 58. This may include, among other things, user feedback expressing “a desired length of the transcript summary.” Chalana ¶ 30.
compute a reduced plurality of thought objects, the reduced plurality of thought objects comprising at least a portion of the plurality of thought objects;
“At decision block 420, audio-visual media synopsis module 400 may determine whether user input or feedback has been received indicating a change to transcript 335, transcript summary 340, or video summary 345. The user feedback may comprise an instruction to include or not include a portion of transcript 335 . . . in (a potentially re-rendered) video summary 345.” Chalana ¶ 65; see also Chalana ¶ 30 (“User feedback to audio-visual media synopsis module 400 may comprise, for example . . . an instruction to include or exclude a portion of the audio-visual media from at least one of the transcript [or] the transcript summary.”)
Then, audio-visual media synopsis module 400 implements the changes by excluding or including selected portions of the transcript, and inputting the modified version transcript 335 into transcription summary module 600, “with or without the included or excluded portions.” Chalana ¶ 66.
generate a plurality of clusters by analyzing a semantic vector representation of thought objects in the reduced plurality of thought objects
Having called transcription summarization module 600 with the modified transcript 335 as input, “transcription summarization module 600 may receive an input transcription, such as from audio-visual media transcription module 500, such as transcript 335,” Chalana ¶ 79, “convert transcript 335 and video 305 into plurality of vectors or tensors,” Chalana ¶ 80, and then process the vectors or tensors to determine a plurality of “sentence meaning clusters in the transcript.” Chalana ¶ 83. Accordingly, “[a]t block 620, transcription summarization module 600 may output sentence meaning clusters.” Chalana ¶ 84.
wherein a quantity of clusters in the plurality of clusters is based on the length of the summary
“A number of such sentence meaning clusters may be determined based on a desired length of a summary, such as from block 410.” Chalana ¶ 84.
generate a summary of the plurality of text segments, the summary based on one or more thought objects from the plurality of clusters;
“For each sentence meaning cluster or topic, at block 625, transcription summarization module 600 may select a number of sentences from each cluster, for example, one or more sentences from each sentence meaning cluster.” Chalana ¶ 86. Then, “[a]t block 630, transcription summarization module 600 may output transcript summary 340 comprising, for example, sentences selected at block 625.” Chalana ¶ 88.
and display the summary on a graphical user interface.
As shown in FIG. 9, audio-visual media synopsis module 400 displays the transcript summary 340 it directed transcription summarization module 600 to generate in the earlier steps, within “transcript summary window 935.” Chalana ¶ 122.
Chalana does not appear to explicitly consider “a quantity of thought objects comprised within the reduced plurality of thought objects” as a factor when generating its plurality of sentence meaning clusters.
Liu, however, teaches a technique for summarizing text that considers both this factor and desired length. Much like the claimed invention, the technique involves the following:
receive a length of a summary;
In Liu’s summarization algorithm, “the summary length [is] fixed by the user.” Liu 2594.
generate a plurality of clusters by analyzing a semantic vector representation of thought objects
Each sentence in the input document collection is converted into a vector within a vector space model (“VSM”), Liu 2593, and then the vectors are clustered using “cosine similarity and k-means clustering.” Liu 2594.
wherein a quantity of clusters in the plurality of clusters is based on the length of the summary and a quantity of thought objects comprised within the reduced plurality of thought objects;
Liu proposes multiple methods for selecting the “k” number of clusters to be employed during k-means clustering, one of which involves using “the length of summary divided by the average length of sentences in document collection” as “k.” Liu 2594. The average length of the sentences in the document collection is necessarily based on the total number of sentences, because that is part of the definition of “average” (i.e., the sum of a set of values divided by the quantity of values—in this case, the sum of each sentence’s length divided by the quantity of sentences).
In other words, Liu chooses the number of clusters based on (1) the desired summary length, (2) the total length of all of the sentences, and (3) the number sentences. Since the claim only requires (1) and (3), Liu teaches both elements of this limitation.
generate a summary of the plurality of text segments, the summary based on one or more thought objects from the plurality of clusters;
“For each sentence cluster, we need to select one sentence to represent the topic denoted by the cluster.” Liu 2594.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to improve Chalan’s text summarization process with the same technique that Liu used to improve similar methods of text summarization, i.e., by setting the number of sentence clusters used for summarization as the desired summary length divided by the average sentence length of the source text. One would have been motivated to use Liu’s strategy because “[t]he most probable number of sentences in a fixed-length-summary is the length of summary divided by the average length of sentences in document collection.” Liu 2594.
Claim 4
Chalana and Liu teach the system of claim 1, wherein to generate the plurality of clusters, the plurality of programming instructions when executed by the processor, further cause the processor to
associate a cluster, of the plurality of clusters, to at least a portion of the reduced plurality of thought objects.
“At block 620, transcription summarization module 600 may output sentence meaning clusters. Semantic meaning of sentences within each cluster may be similar; semantic meaning of sentences between clusters may be different.” Chalana ¶ 84.
Claim 5
Chalana and Liu teach the system of claim 1, wherein the plurality of programming instructions when executed by the processor, further cause the processor to
select the one or more thought objects from each of the generated clusters using one or more confidence scores, a thought object theming process, a thought object sentiment analysis process, and/or a thought object rating process, or a combination thereof.
“For each sentence meaning cluster or topic, at block 625, transcription summarization module 600 may select a number of sentences from each cluster, for example, one or more sentences from each sentence meaning cluster. The selected sentence may be a highest ranking sentence in a cluster, a highest ranking sentence which is most dissimilar from sentences in other clusters, a fixed number, a number to achieve a percentage of video 305, and the like.” Chalana ¶ 86. “In embodiments, ranking may not be cluster oriented, but may, for example, be based on text-rank and/or on text-rank including a component of document frequency, including inverse document frequency.” Chalana ¶ 84.
Liu, in an overlapping teaching, provides several additional strategies for selecting representative sentences from each of the clusters, which fall under one or more of the claimed processes. See Liu 2594–2595.
Claim 6
Chalana and Liu teach the system of claim 5,
wherein the one or more confidence scores are indicative of a quantified importance of each thought object.
“For each sentence meaning cluster or topic, at block 625, transcription summarization module 600 may select a number of sentences from each cluster, for example, one or more sentences from each sentence meaning cluster. The selected sentence may be a highest ranking sentence in a cluster, a highest ranking sentence which is most dissimilar from sentences in other clusters, a fixed number, a number to achieve a percentage of video 305, and the like.” Chalana ¶ 86. “In embodiments, ranking may not be cluster oriented, but may, for example, be based on text-rank and/or on text-rank including a component of document frequency, including inverse document frequency.” Chalana ¶ 84.
Claim 10
Chalana and Liu teach the system of claim 1, wherein the plurality of programming instructions when executed by the processor, further cause the processor to
generate a pre-configured number of clusters.
Chalana teaches that its transcription summarization module 600 is pre-programmed to create a number of clusters that accords with the desired length of the summary. Chalana ¶ 84.
Claims 11, 14–16, and 20
Claims 11, 14–16, and 20 are rejected for reciting a method that includes all of the steps that the respective systems of claims 1, 4–6, and 10 perform as part of their normal operation, given that the prior art teaches each and every step performed by those systems.
II. Chalana, Liu, and Baker teach claims 2 and 12.
Claims 2 and 12 are rejected under 35 U.S.C. § 103 as being unpatentable over Chalana in view of Liu as applied to claims 1 and 11 above, and further in view of U.S. Patent Application Publication No. 2015/0339288 A1 (“Baker”).
Claim 2
Chalan and Liu teach the system of claim 1, wherein to compute the reduced plurality of thought objects the plurality of programming instructions when executed by the processor, further cause the processor to:
identify and remove one or more redundant thought objects from the plurality of thought objects
“User feedback to audio-visual media synopsis module 400 may comprise, for example . . . an instruction to include or exclude a portion of the audio-visual media from at least one of the transcript [or] the transcript summary.” Chalana ¶ 30.
Chalana and Liu do not appear to explicitly disclose identifying the redundant thought objects for removal “based on information associated with the plurality of thought objects,” since Chalana relies on the user to manually say which sentences should be removed from the transcript.
Baker, however, teaches a text summarization system 110 that was improved in the same way that the invention of claim 2 was improved over Chalana and Liu’s teachings, i.e., wherein to compute the reduced plurality of thought objects the plurality of programming instructions when executed by the processor, further cause the processor to:
identify and remove one or more redundant thought objects from the plurality of thought objects, the redundant thought objects identified based on information associated with the plurality of thought objects.
Summarization system 110 includes programming instructions that reduce the total amount of text to be processed for summarization in a few different ways. For one, “extraction module 206 [] removes boilerplate text snippets that do not belong to the body of text of the reported story.” Baker ¶ 27. Second, a “deduplication module 210” discards all of the text belonging to RSS articles deemed to be duplicates, e.g., articles from two different publishers concerning the same news story, or republications of the same article. Baker ¶¶ 34–35. Third, a “preprocessing module 212” removes all of the “stopwords” from the collected text, such as “a”, “about”, “after”, “because”, “between”, “the”, “for”, “or”, etc. Baker ¶ 30.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to improve Chalana and Liu’s summarization systems in the same way that Baker improved his summarization system 110, i.e., by removing boilerplate text and duplicate articles prior to performing summarization. One would have been motivated to improve Chalana and Liu with Baker’s technique because Baker’s technique automates an otherwise manual activity. Automating a known manual activity, when the means for automation is already known, is a well-established rationale for combining references. See MPEP § 2144.04.
Claim 12
Claim 12 is rejected according to the same findings and rationale as provided above for claims 11 and 2, combined.
III. Chalana, Liu, and Tian teach claims 3 and 13.
Claims 3 and 13 are rejected under 35 U.S.C. § 103 as being unpatentable over Chalana and Liu as applied to claims 1 and 11 above, and further in view of Chinese Patent Application Publication No. 113806534 A (“Tian”).
Claim 3
Chalana and Liu teach the system of claim 1, but do not explicitly disclose “responsive to the plurality of thought objects being above a first threshold, generate a random sample from the plurality of the thought objects.”
Tian, however, teaches an improvement technique (“step 1.1.1” of FIG. 2) where, when a source data set is “too large,” the process uses only a random sampling of the data, rather than the entire data set. See Tian FIG. 2 and accompanying translation of step 1.1.1.
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to improve Chalana and Liu with Tian’s technique of randomly sampling the data when it is too large, rather than processing the entire set of sentences. One would have been motivated to improve Chalana and Liu with Tian’s technique because random sampling provides a way to reduce processing loads without sacrificing too much by way of accuracy (since the sample is likely to be representative of the overall dataset).
Claim 13
Claim 13 is rejected according to the same findings and rationale as provided above for claims 11 and 3, combined.
Conclusion
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 Justin R. Blaufeld whose telephone number is (571)272-4372. The examiner can normally be reached M-F 9:00am - 4:00pm ET.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, James K Trujillo can be reached at (571) 272-3677. 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.
Justin R. Blaufeld
Primary Examiner
Art Unit 2151
/Justin R. Blaufeld/Primary Examiner, Art Unit 2151
Prosecution Insights