DETAILED ACTION
Notice of 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 .
Claim Rejections - 35 USC § 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 an abstract idea without significantly more.
Regarding Step 1 of the Alice/Mayo framework, Claims 1-8 are directed to a method (a process), Claims 9-16 are directed to a system (a machine), and Claims 17-20 are directed to a non-transitory computer-readable device (an article of manufacture), which each fall within one of the four statutory categories of inventions.
Regarding Claim 1
Step 2A, prong 1 (Is the claim directed to a law of nature, a natural phenomenon or an abstract idea).
Claim 1 recites the following mental processes, that in each case under the broadest reasonable interpretation, covers performance of the limitation in the mind (including an observation, evaluation, judgment, opinion) or with the aid of pencil and paper but for the recitation of generic computer components (e.g., “computer-implemented”, “sentiment classification model trained”).
preprocessing the data to generate preprocessed data; (under the broadest reasonable interpretation, a human can mentally review a call transcript and preprocess the transcript, e.g., to correct misspellings on paper using a pen)
applying the preprocessed data ... to produce a plurality of probability values respectively corresponding to a plurality of sentiment categories corresponding to different potential qualities of the call transcript text; and (under the broadest reasonable interpretation, a human can mentally review a call transcript and, for each sentence, estimate a probability that such sentence has a negative, neutral, or positive sentiment, for multiple qualities of the transcript, such as for both agent and customer utterances)
determining a numeric sentiment classification based on comparing the plurality of probability values to a plurality of thresholds respectively corresponding to the plurality of sentiment categories, wherein the numeric sentiment classification corresponds to a quality of the call. (under the broadest reasonable interpretation, a human can mentally determine a numeric sentiment polarity, such as between -1 to 1, where -1 is a strong negative polarity, 0 is a neutral polarity, and +1 is a strong positive polarity, and where there may be thresholds at -0.5, -.1, .1, and .5)
Step 2A, prong 2 (Does the claim recite additional elements that integrate the judicial exception into a practical application?).
The judicial exception is not integrated into a practical application. In particular, the claim recites the additional elements (e.g., “computer-implemented” and “graph neural network”) which are recited at a high-level of generality such that they amount to no more than mere instructions to apply the exception using a generic computer component (See MPEP 2106.05(f)).
Regarding the “receiving data including call transcript text and metadata related to a call” limitation, such additional element of a data gathering step is recited at a high level of generality and amounts to extra-solution activity of receiving data, i.e. pre-solution activity of gathering data for use in the claimed process (see MPEP 2106.05(g)).
Regarding the “to a sentiment classification model trained” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception. In particular, the claim only recites the additional element of a generic trained model. This additional element is recited at a high-level of generality and amounts to no more than mere instructions to apply the exception using a generic computer component (a generic trained model). Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea (See MPEP 2106.05(f)).
Accordingly, at Step 2A, prong two, after considering all claim elements individually and as an ordered combination, it is determined that the claims do not integrate the judicial exception into a practical application.
Step 2B (Does the claim recite additional elements that amount to significantly more than the judicial exception?)
In accordance with Step 2B, the claim does not include additional elements that are sufficient to amount to significantly more that the judicial exception. As discussed above, the additional elements (e.g., “computer-implemented” and “graph neural network”) are recited at a high-level of generality such that they amount to no more than mere instructions to apply the exception using a generic computer component (See MPEP 2106.05(f)).
Regarding the “receiving data including call transcript text and metadata related to a call” limitation, as discussed above, the additional element of a data gathering step is recited at a high level of generality and amounts to extra-solution activity of receiving data, i.e. pre-solution activity of gathering data for use in the claimed process. The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory").
Regarding the “to a sentiment classification model trained” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception, because the limitation merely provides instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. Accordingly, this additional element does not add significantly more than the judicial exception. (See MPEP 2106.05(f)).
Accordingly, at Step 2B after considering all claim elements individually and as an ordered combination, it is determined that the claims do not integrate the judicial exception into a practical application.
Regarding Claim 2
Step 2A, Prong 1
wherein the preprocessing further comprises: removing call transcript text corresponding to a customer care agent from the data. (under the broadest reasonable interpretation, a human can mentally review call transcript text and on the paper transcript, cross-out all of the customer care agent utterances from the transcript)
Regarding Step 2A, Prong 2, the claim does not include any additional elements that integrate the judicial exception into a practical application and regarding Step 2B, there are no additional elements recited that amount to significantly more than the judicial exception.
Regarding Claim 3
Step 2A, Prong 1
wherein the preprocessing further comprises: replacing one or more similar words from the call transcript text with a predetermined word. (under the broadest reasonable interpretation, a human can mentally replace words in the transcript, such as scratching out every instance of a month being spelled out and writing in [MONTH])
Regarding Step 2A, Prong 2, the claim does not include any additional elements that integrate the judicial exception into a practical application and regarding Step 2B, there are no additional elements recited that amount to significantly more than the judicial exception.
Regarding Claim 4
Step 2A, Prong 1
wherein the plurality of sentiment categories include a positive numerical category, a negative numerical category, and a neutral category. (under the broadest reasonable interpretation, a human can mentally determine a numeric sentiment polarity, such as between -1 to 1, where -1 is a strong negative polarity, 0 is a neutral polarity, and +1 is a strong positive polarity)
Regarding Step 2A, Prong 2, the claim does not include any additional elements that integrate the judicial exception into a practical application and regarding Step 2B, there are no additional elements recited that amount to significantly more than the judicial exception.
Regarding Claim 5
Step 2A, Prong 2
Regarding the “wherein the metadata includes crosstalk identification” limitation, such additional element of a data gathering step is recited at a high level of generality and amounts to extra-solution activity of receiving data, i.e. pre-solution activity of gathering data for use in the claimed process (see MPEP 2106.05(g)).
Step 2B
Regarding the “wherein the metadata includes crosstalk identification” limitation, as discussed above, the additional element of a data gathering step is recited at a high level of generality and amounts to extra-solution activity of receiving data, i.e. pre-solution activity of gathering data for use in the claimed process. The courts have found limitations directed to obtaining information electronically, recited at a high level of generality, to be well-understood, routine, and conventional (see MPEP 2106.05(d)(II), “receiving or transmitting data over a network”, "electronic record keeping," and "storing and retrieving information in memory").
Regarding Claim 6
Step 2A, Prong 1
wherein the applying further comprises: extracting a plurality of text features from the preprocessed data. (under the broadest reasonable interpretation, a human can mentally extract text features, such as a word count for each of the speakers, average number of words per utterance for each speaker, etc.)
Regarding Step 2A, Prong 2, the claim does not include any additional elements that integrate the judicial exception into a practical application and regarding Step 2B, there are no additional elements recited that amount to significantly more than the judicial exception.
Regarding Claim 7
Step 2A, Prong 1
wherein the plurality of text features are weighted based on relative importance (under the broadest reasonable interpretation, a human can mentally weight certain text features more than others, such as mentally weighting utterances by the customer higher than utterances by the customer service representative)
Step 2A, Prong 2
Regarding the “by the sentiment classification model” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception. In particular, the claim only recites the additional element of a generic model. This additional element is recited at a high-level of generality and amounts to no more than mere instructions to apply the exception using a generic computer component (a generic model). Accordingly, this additional element does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea (See MPEP 2106.05(f)).
Step 2B
Regarding the “by the sentiment classification model” limitation, such limitation is recited at a high-level of generality and amounts to no more than adding the words “apply it” (or an equivalent) with the judicial exception, because the limitation merely provides instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea. Accordingly, this additional element does not add significantly more than the judicial exception. (See MPEP 2106.05(f)).
Regarding Claim 8
Step 2A, Prong 1
separating the call transcript text into a plurality of portions corresponding to a plurality of customer care agents. (under the broadest reasonable interpretation, a human can separate the call transcript text according to different customer care agents, such as by noting care agent 1 and care agent 2 with a different colored pen)
Regarding Step 2A, Prong 2, the claim does not include any additional elements that integrate the judicial exception into a practical application and regarding Step 2B, there are no additional elements recited that amount to significantly more than the judicial exception.
Regarding Claim 9
Step 2A, Prong 1
Claim 9 recites a system that corresponds to the method of claim 1, and therefore the analysis under Step 2A, Prong 1 with respect to claim 1 also applies to this claim 9. While claim 9 recites additional generic computing components (“processor”, “memory”, “sentiment classification model trained”), such additional generic computing components do not change the analysis under Step 2A, Prong 1.
Step 2A, Prong 2
Claim 9 recites a system that corresponds to the method of claim 1, and therefore the analysis under Step 2A, Prong 2 with respect to claim 1 also applies to this claim 9. While claim 9 recites additional generic computing components (“processor”, “memory”, “sentiment classification model trained”), such additional generic computing components do not change the analysis under Step 2A, Prong 2.
Step 2B
Claim 9 recites a system that corresponds to the method of claim 1, and therefore the analysis under Step 2B with respect to claim 1 also applies to this claim 9. While claim 9 recites additional generic computing components (“processor”, “memory”, “sentiment classification model trained”), such additional generic computing components do not change the analysis under Step 2B.
Claims 10-16 each depend from claim 9 and each correspond to the method of claims 2-8, respectively, and are therefore rejected for the same reasons explained above with respect to claim 9 and claims 2-8, respectively.
Regarding Claim 17
Step 2A, Prong 1
Claim 17 recites a non-transitory computer-readable device that corresponds to the method of claim 1, and therefore the analysis under Step 2A, Prong 1 with respect to claim 1 also applies to this claim 17. While claim 17 recites additional generic computing components (“non-transitory computer-readable device”, “computing device”, “sentiment classification model trained”), such additional generic computing components do not change the analysis under Step 2A, Prong 1.
Step 2A, Prong 2
Claim 17 recites a non-transitory computer-readable device that corresponds to the method of claim 1, and therefore the analysis under Step 2A, Prong 2 with respect to claim 1 also applies to this claim 17. While claim 17 recites additional generic computing components (“non-transitory computer-readable device”, “computing device”, “sentiment classification model trained”), such additional generic computing components do not change the analysis under Step 2A, Prong 2.
Step 2B
Claim 17 recites a non-transitory computer-readable device that corresponds to the method of claim 1, and therefore the analysis under Step 2B with respect to claim 1 also applies to this claim 17. While claim 17 recites additional generic computing components (“non-transitory computer-readable device”, “computing device”, “sentiment classification model trained”), such additional generic computing components do not change the analysis under Step 2B.
Claims 18-20 each depend from claim 17 and each correspond to the method of claims 2-4, respectively, and are therefore rejected for the same reasons explained above with respect to claim 17 and claims 2-4, respectively.
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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1, 4, 6, 8-9, 12, 14, 16-17, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over US 20210342554 A1, hereinafter referenced as MARTIN, in view of US 20120254063 A1, hereinafter referenced as RITTERMAN.
Regarding Claim 1
MARTIN teaches:
A computer implemented method, comprising: (MARTIN, para. 0095: “Methods discussed above may be implemented by hardware, software, firmware, middleware, microcode, hardware description languages, or any combination thereof.”)
receiving data including call transcript text and metadata related to a call; (MARTIN, para. 0028: “Transactions can include interactions with a customer, e.g., such as a call-center call, an instant messaging chat, a teleconference, a video conference, a survey, an email exchange with a customer, etc.”;
MARTIN, para. 0030: “The server 110 may have access to data sources 180. Data sources 180 may be part of the server 110. Data sources 180 may be remote from, but accessible by the server 110. Data sources 180 be stored at a third-party server 190 and shared with the server 110. Data sources 180 may represent data from various sources, e.g., various social media platforms, data from a call center, data from an email server, data from a service, data from a news feed, etc. Thus, data sources 180 may represent very different types of sources, some of which may be remote and some of which may be local to server 110.”
MARTIN, para. 0031: “The pre-processing system 112 may optionally be used to prepare the data from the data sources 180 for processing. For example, for data sources originating from an audio or video source (e.g., a call center, a short video recording in a social media post, etc.), the data may be transcribed into text.”
MARTIN, para. 0081: “The transactions are received from one or more data sources, e.g., data sources 180 of FIG. 1. Some of the transactions may have been pre-processed prior to being received by the system. For example, if the transaction is an audio call or video call, the transaction may have been processed by a transcription service to convert the conversation in the call to text. If the transaction did arise from an audio or video file and has not already been processed by a transcription service, the system may process the transaction with a transcription service (901). The transcription service converts audio dialog into text and sentences (where possible). Some transcription services also label the sentences with a speaker. Sometimes the speaker is identified by a name or phone number, for example if such information was provided prior to joining a conference call. Sometimes a speaker may be identified by a generic number or identifier (e.g., Speaker 1). Text-based contacts, such as a chat session or an email chain, do not need pre-processing with a transcription service.”
Examiner’s Note: the server 120 receives call-center data 180, which can be transcribed using a transcription service, where sentences are labeled with a speaker name/phone-number (corresponding to recited “metadata”))
preprocessing the data to generate preprocessed data; (MARTIN, para. 0031: “The pre-processing system 112 may optionally be used to prepare the data from the data sources 180 for processing. For example, for data sources originating from an audio or video source (e.g., a call center, a short video recording in a social media post, etc.), the data may be transcribed into text.”)
applying the preprocessed data to a sentiment classification model trained to produce a plurality of
(MARTIN, para. 0035: “The system classifiers may include machine-learned classifiers. Example classifiers include a sentiment classifier, a reason detector, an intent detector, effort scoring, an emotion detector, an emotional intensity classifier, an empathy detector, an issue resolution detector, a topic detector, span detection, a profanity detector, a named entity recognizer, an emotional intelligence classifier, etc.”
MARTIN, para. 0043: “In this example, the scoring units are sentences from the call. The scoring units are each associated with one of two verbatims, Agent or Customer. Some classifiers may only apply to certain verbatims, e.g., an empathy detector may only be applied to the Agent's verbatims. Even if a classifier is applied to every scoring unit, some summary templates may only use the score from scoring units for certain speakers, e.g., a sentiment classifier may be applied to all scoring units but a template variable may use only sentiment from verbatims from the customer.”
MARTIN, para. 0045: “The sentiment classifier 255 may tag a scoring unit with one or more sentiment tags. In some implementations, a sentiment tag may indicate positive sentiment or negative sentiment.”
MARTIN, para. 0066: “Customer Jane Park called about late delivery. Agent Smith gave the Miranda Warning. Agent Smith offered a future credit. Customer's sentiment was positive and customer expressed satisfaction. Issues are resolved. Agent rubric score: 96.”;
Examiner’s Note: sentences from the pre-processed transcript data are provided to a machine-learning trained sentiment classifier (corresponding to recited “sentiment classification model trained to produce...”), where a positive/negative sentiment is classified (corresponding to the recited “plurality of sentiment categories”, where positive is 1 category and negative is another category), and where sentiments can be for each individual speaker on the call (corresponding to recited “different potential qualities of the call transcript text”)
However, MARTIN fails to explicitly teach:
probability values respectively corresponding to a plurality of sentiment categories
determining a numeric sentiment classification based on comparing the plurality of probability values to a plurality of thresholds respectively corresponding to the plurality of sentiment categories, wherein the numeric sentiment classification corresponds to a quality of the call.
However, in a related field of endeavor (determining sentiment levels from natural language information, see para. 0027), RITTERMAN teaches and makes obvious:
probability values respectively corresponding to a plurality of sentiment categories (RITTERMAN, para. 0052: “At block 450, MRS 100 may use the number of intersections to calculate a probability that the word indicates positive or negative sentiment. For example, at block 420 the word "abysmal" from input string 421 may be identified for further processing. At block 430, it may be determined the word "abysmal" has fifteen synonyms. At block 440, it may be determined that, of the fifteen synonyms, twelve synonyms are listed in the list of negative synonyms obtained at block 410, and one synonym is listed in the list of positive synonyms obtained at block 410. Thus, at block 450, it may be determined that there is an 80% probability that "abysmal" indicates negative sentiment, and a 7% probability that "abysmal" indicates positive sentiment. In one embodiment, MRS 100 may consider the word "abysmal" to indicate negative sentiment, because the probability of "abysmal" indicating a negative sentiment is greater than the probability of it indicating a positive sentiment, and/or the probability of "abysmal" indicating a negative sentiment is greater than a threshold value (e.g., 50%).”;
Examiner’s Note: RITTERMAN discloses that the output of a sentiment classifier can be a probability that a word unit indicates positive or negative sentiment; the MARTIN-RITTERMAN combination now modifies the sentiment classifier of MARTIN to output a percentage probability associated with positive/negative sentiment as in RITTERMAN)
determining a numeric sentiment classification based on comparing the plurality of probability values to a plurality of thresholds respectively corresponding to the plurality of sentiment categories, wherein the numeric sentiment classification corresponds to a quality of the call. (RITTERMAN, para. 0040: “At block 220, MRS 100 may determine a level of sentiment that is represented by the incoming data. Possible levels of sentiments include positive, negative, and neutral sentiment. It should be noted, however, that additional levels of sentiments, such as very positive, slightly positive, and the like, are possible. In addition, in some embodiments, sentiment levels may be indicated using a numerical value. For example, a value of +1.0 may be used to indicate strong positive sentiment, while a value of -1.0 may be used to indicate strong negative sentiment.”
RITTERMAN, para. 0053: “It should be noted that other thresholds may be used, and that a word may be considered neutral if the probability of the word as being indicative of positive or negative sentiment are both less than a certain threshold (e.g., 50%). Neutral words are potentially informative, because they can indicate the market adoption of a product because the use of a product name in a casual comment indicates public brand awareness.”;
Examiner’s Note: RITTERMAN discloses that a sentiment can be numeric, such as +1.0 for positive sentiment and -1.0 for negative sentiment, and further discloses having at least 2 different thresholds (one for positive sentiment and one for negative sentiment, otherwise its neutral sentiment); the MARTIN-RITTERMAN combination now modifies the sentiment classifier of MARTIN to output a percentage probability associated with positive/negative sentiment expressed numerically as in RITTERMAN, where such numeric sentiment can be mapped to positive/neutral/negative categories based on thresholds as disclosed by RITTERMAN)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of MARTIN and RITTERMAN as explained above. As disclosed by RITTERMAN, one of ordinary skill would have been motivated to do so in order to include polarity into the sentiment, so it could be strongly positive or strongly negative. (para. 0040). One of ordinary skill would further have been motivated to do so in order to associate probabilities with the sentiment category, to provide further insight into how strong the prediction is.
Regarding Claim 4
MARTIN and RITTERMAN disclose the method of claim 1. However, MARTIN fails to explicitly teach:
wherein the plurality of sentiment categories include a positive numerical category, a negative numerical category, and a neutral category. (RITTERMAN, para. 0040: “At block 220, MRS 100 may determine a level of sentiment that is represented by the incoming data. Possible levels of sentiments include positive, negative, and neutral sentiment. It should be noted, however, that additional levels of sentiments, such as very positive, slightly positive, and the like, are possible. In addition, in some embodiments, sentiment levels may be indicated using a numerical value. For example, a value of +1.0 may be used to indicate strong positive sentiment, while a value of -1.0 may be used to indicate strong negative sentiment.”;
Examiner’s Note: RITTERMAN discloses that a sentiment can be numeric, such as +1.0 for positive sentiment and -1.0 for negative sentiment, and further discloses having neutral sentiment; the MARTIN-RITTERMAN combination now modifies the sentiment classifier of MARTIN to output positive/negative sentiment expressed numerically as in RITTERMAN, where such numeric sentiment can be mapped to positive/neutral/negative categories based on thresholds as disclosed by RITTERMAN)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of MARTIN and RITTERMAN as explained above. As disclosed by RITTERMAN, one of ordinary skill would have been motivated to do so in order to include polarity into the sentiment, so it could be strongly positive or strongly negative. (para. 0040). One of ordinary skill would further have been motivated to do so in order to associate probabilities with the sentiment category, to provide further insight into how strong the prediction is.
Regarding Claim 6
MARTIN and RITTERMAN disclose the method of claim 1. MARTIN further teaches:
wherein the applying further comprises: extracting a plurality of text features from the preprocessed data. (MARTIN, para. 0033: “For example, the resegmentation service 124 may combine the two sentences “okay. I got it.” into “okay, i got it”, including updating the metadata (attributes) of the sentences calculated during preprocessing.”;
MARTIN, para. 0080: “The metadata generated by process 900 can be used as features that support various analyses tasks, e.g., supporting one or more system classifiers (e.g., classifiers 136).”
MARTIN, para. 0093: “The sequencing metadata may be added before the scoring units are labeled or after the scoring units are labeled. The sequencing metadata includes scoring unit, participant, and transaction level metadata. The system may then generate features describing the labeled transactions (1110). For example, each transaction may have a vector with several hundred or more dimensions, with each dimension representing a different feature. The features (dimensions) are based on the content of the scoring units as well as the metadata describing the scoring units, any special events, the sequencing of the scoring units and special events, the participants, the transaction, etc.”)
Regarding Claim 8
MARTIN and RITTERMAN disclose the method of claim 1. MARTIN further teaches:
separating the call transcript text into a plurality of portions corresponding to a plurality of customer care agents. (MARTIN, para. 0036: “Thus, for example, if a customer service call was transferred, a call transfer template may be inserted and evaluated so that the template variables in the call transfer template are replaced according to their respective applicable variable replacement logic.”
MARTIN, para. 0081: “The transcription service converts audio dialog into text and sentences (where possible). Some transcription services also label the sentences with a speaker. Sometimes the speaker is identified by a name or phone number, for example if such information was provided prior to joining a conference call. Sometimes a speaker may be identified by a generic number or identifier (e.g., Speaker 1). Text-based contacts, such as a chat session or an email chain, do not need pre-processing with a transcription service.”;
Examiner’s Note: If a customer service call is transferred (such as to a higher level service rep), the transcript would have more than one customer service rep identified as a speaker in the transcript)
Regarding Claim 9
MARTIN teaches:
A system, comprising: a memory; and at least one processor coupled to the memory and configured to: (MARTIN, para. 0106: “In one aspect, a system for includes at least one processor, memory storing a library of classifiers, and memory storing instructions that, when executed by the at least one processor, causes the system to”)
The remaining limitations correspond to the method of claim 1 and are therefore rejected for the same reasons explained above with respect to claim 1.
Claim 12 depends from claim 9 and recites a system that corresponds to the method of claim 4, and is therefore rejected for the same reasons explained above with respect to claims 4 and 9.
Claim 14 depends from claim 9 and recites a system that corresponds to the method of claim 6, and is therefore rejected for the same reasons explained above with respect to claims 6 and 9.
Claim 16 depends from claim 9 and recites a system that corresponds to the method of claim 8, and is therefore rejected for the same reasons explained above with respect to claims 8 and 9.
Regarding Claim 17
MARTIN teaches:
A non-transitory computer-readable device having instructions stored thereon that, when executed by at least one computing device, cause the at least one computing device to perform operations comprising: (MARTIN, para. 0106: “In one aspect, a system for includes at least one processor, memory storing a library of classifiers, and memory storing instructions that, when executed by the at least one processor, causes the system to”)
The remaining limitations correspond to the method of claim 1 and are therefore rejected for the same reasons explained above with respect to claim 1.
Claim 20 depends from claim 17 and recites a non-transitory computer-readable device that corresponds to the method of claim 4, and is therefore rejected for the same reasons explained above with respect to claims 4 and 17.
Claims 2, 10, and 18 are rejected under 35 U.S.C. 103 as being unpatentable over MARTIN in view of RITTERMAN and further in view of US 11487948 B2, hereinafter referenced KAUR.
Regarding Claim 2
MARTIN and RITTERMAN disclose the method of claim 1. However, MARTIN and RITTERMAN fail to explicitly teach:
wherein the preprocessing further comprises: removing call transcript text corresponding to a customer care agent from the data.
However, in a related field of endeavor (AI dialog systems, see col. 1, lines 6-8), KAUR teaches and makes obvious:
wherein the preprocessing further comprises: removing call transcript text corresponding to a customer care agent from the data. (KAUR, col. 6, lines 31-34: The text-based chat corpus may be pre-processed, and customer utterances and agent utterances are preferably separated and accompanied by time stamps.”;
KAUR, claim 1: “separating customer utterances from agent utterances and operating only on the customer utterances by extracting customer utterances immediately following an agent signature utterance, searching for intent indicator phrases within the extracted customer utterances, preprocessing the customer utterances including the intent indicator phrases, and outputting the pre-processed customer utterances as the identified relevant utterances”;
Examiner’s Note: KAUR teaches separating agent and customer utterances, so that only customer utterances are selected for analysis; the MARTIN-RITTERMAN-KAUR combination now modifies the system of MARTIN to separate customer utterances from agent utterances as in KAUR).
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of MARTIN, RITTERMAN, and KAUR as explained above. As disclosed by KAUR, one of ordinary skill would have been motivated to do so in order to focus only on the customer intent, and therefore ignoring less-relevant agent transcript data. (col. 9, lines 21-27).
Claim 10 depends from claim 9 and recites a system that corresponds to the method of claim 2, and is therefore rejected for the same reasons explained above with respect to claims 2 and 9.
Claim 18 depends from claim 17 and recites a non-transitory computer-readable device that corresponds to the method of claim 2, and is therefore rejected for the same reasons explained above with respect to claims 2 and 17.
Claims 3, 11, and 19 are rejected under 35 U.S.C. 103 as being unpatentable over MARTIN in view of RITTERMAN and further in view of US 20190013012 A1, hereinafter referenced as HWANG.
Regarding Claim 3
MARTIN and RITTERMAN disclose the method of claim 1. However, MARTIN and RITTERMAN fail to explicitly teach:
wherein the preprocessing further comprises: replacing one or more similar words from the call transcript text with a predetermined word.
However, in a related field of endeavor (text processing), HWANG teaches and makes obvious:
wherein the preprocessing further comprises: replacing one or more similar words from the call transcript text with a predetermined word. (HWANG, para. 0037: “The corpus enhancing unit 110 may obtain a word having a similarity of a predetermined level or greater with a word that is includes in the basis sentence corpus by performing word embedding or paraphrase, and enhance the basis sentence corpus by using the obtained word. In detail, the corpus enhancing unit 110 may generate a new sentence by replacing a word or noun which is included in a basis sentence with a synonym, and thus enhance the basis sentence corpus.”;
HWANG, para. 0047: “When words having a similarity of a predetermined level or greater with a word constituting basis sentences are obtained, in step S203, the corpus enhancing unit 110 may replace a word constituting the basis sentence with the obtained similar word, obtain a new basis sentence, and thus enhance the basis sentence corpus on the basis of the same.”
Examiner’s Note: the MARTIN-RITTERMAN-HWANG combination now modifies MARTIN to replace a similar word in a call transcript with an obtained word from a corpus as in HWANG, such as to generate additional samples for machine learning training)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of MARTIN, RITTERMAN, and HWANG as explained above. As disclosed by HWANG, one of ordinary skill would have been motivated to do so in order to enhance the size of a training corpus for unsupervised machine learning. (para. 0009).
Claim 11 depends from claim 9 and recites a system that corresponds to the method of claim 3, and is therefore rejected for the same reasons explained above with respect to claims 3 and 9.
Claim 19 depends from claim 17 and recites a non-transitory computer-readable device that corresponds to the method of claim 3, and is therefore rejected for the same reasons explained above with respect to claims 3 and 17.
Claims 5 and 13 are rejected under 35 U.S.C. 103 as being unpatentable over MARTIN in view of RITTERMAN and further in view of US 20210157834 A1, hereinafter referenced as SIVASUBRAMANIAN.
Regarding Claim 5
MARTIN and RITTERMAN disclose the method of claim 1. However, MARTIN and RITTERMAN fail to explicitly teach:
wherein the metadata includes crosstalk identification.
However, in a related field of endeavor (call center analytics, see para. 0002), SIVASUBRAMANIAN teaches and makes obvious:
wherein the metadata includes crosstalk identification. (SIVASUBRAMANIAN, para. 0042: “Call audio and transcripts may be provided together with additional metadata associated with the call, such as sentiment scoring for different segments of a call. A contact search page may be used for conducting fast full-text search on call transcripts. In at least some embodiments, users can filter by entities (e.g., product names), sentiment, and other call characteristics. In some cases, calls are analyzed to extract different call characteristics which may include one or more of the following non-limiting examples: talk speed, interruptions, silence (e.g., gaps in speech), speaker energy, pitch, tone, and other voice characteristics.”
SIVASUBRAMANIAN, para. 0073: “ Speech-to-text service 130 may generate metadata for audio which can include periods of silence, cross-talk (e.g., where multiple speakers talk over each other), and more. Metadata may be included as part of a transcript output.”)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of MARTIN, RITTERMAN, and SIVASUBRAMANIAN as explained above. As disclosed by SIVASUBRAMANIAN, one of ordinary skill would have been motivated to do so in order to track the number of interruptions to “identify potential areas for improvement” for customer service representatives. (para. 0042).
Claim 13 depends from claim 9 and recites a system that corresponds to the method of claim 5, and is therefore rejected for the same reasons explained above with respect to claims 5 and 9.
Claims 7 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over MARTIN in view of RITTERMAN and further in view of US 20190287142 A1, hereinafter referenced as FAN.
Regarding Claim 7
MARTIN and RITTERMAN disclose the method of claim 1. However, MARTIN and RITTERMAN fail to explicitly teach:
wherein the plurality of text features are weighted based on relative importance by the sentiment classification model.
However, in a related field of endeavor (text analysis, including for emotion and sentiment classification, see para. 0022), FAN teaches and makes obvious:
wherein the plurality of text features are weighted based on relative importance by the sentiment classification model. (FAN, para. 0066: “The determination of the degree of importance may be achieved by a sub-network 340 in the learning network 300. The sub-network 340 takes the semantic feature 332 as an input to determine the corresponding weight value that is used to represent a degree of importance 342 of the subset of text items corresponding to the semantic feature 332.”;
Examiner’s Note: the MARTIN-RITTERMAN-FAN combination now weights certain words to be more important for the classification model as in FAN)
Before the effective filing date of the present application, it would have been obvious to one of ordinary skill in the art to combine the teachings of MARTIN, RITTERMAN, and FAN as explained above. As disclosed by FAN, one of ordinary skill would have been motivated to do so in order to consider the “degree of importance” of semantic features to enable more attention to such semantic features. (para. 0045).
Claim 15 depends from claim 9 and recites a system that corresponds to the method of claim 7, and is therefore rejected for the same reasons explained above with respect to claims 7 and 9.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
US 20110208522 A1 (Pereg) “relates to interaction analysis in general, and to detection of sentiment in automated transcriptions, in particular.” (para. 0001).
Li, Bryan, et al. "Acoustic and lexical sentiment analysis for customer service calls." ICASSP 2019-2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2019. Discloses a sentiment recognizer for real-world call center data using lexical cues. (p. 5876, section 1).
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MICHAEL C LEE whose telephone number is (571)272-4933. The examiner can normally be reached M-F 12:00 pm - 8:00 pm 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, Omar Fernandez Rivas can be reached at 571-272-2589. 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.
/MICHAEL C. LEE/Examiner, Art Unit 2128
Prosecution Insights