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 .
Specification
The disclosure is objected to because of the following informalities:
Paragraph [0072] recites "embodimentsmay," which should read "embodiments may."
Paragraph [0073] recites "based on receiving input data that is more likely to be associated with failure data, the semiconductor inference model SIMD may output a value closer to one ('1')" and then recites "based on receiving input data that is more likely to be associated with failure data, the semiconductor inference model SIMD may output a value closer to zero ('0')." The two sentences recite the same condition with opposite outputs; the second sentence apparently should recite "less likely."
Paragraph [0075] recites the inference data IFDT as "a real number ranging from zero ('0') to less than or equal to one ('1')," whereas paragraph [0079] recites "a real number greater than zero ('0') and less than or equal to one ('1')" and paragraph [0092] recites "equal to or greater than zero ('0') and equal to or less than one ('1')." The value range should be described consistently.
Paragraph [0092] recites "the inference 220 module," which should read "the inference module 220."
Paragraph [0092] recites the input values as "X, f, c, K" (lowercase c), whereas the similarity function is thereafter recited as "C" (uppercase); the nomenclature should be made consistent.
Paragraph [0095] recites "until the index kreaches," which should read "until the index k reaches."
Paragraph [0102] recites "may maintain an trained semiconductor language model," which should read "a trained semiconductor language model."
Paragraphs [0115] and [0122] recite "semiconductor equipement," which should read "semiconductor equipment."
Paragraph [0129] recites "generate an summary SMRY," which should read "generate a summary SMRY."
Paragraph [0136] recites "the summarys SMR1 through SMR3," which should read "the summaries SMR1 to SMR3."
Paragraph [0140] recites "are described in detail in detail," a duplicated phrase, and recites "the scope of the disclosure is not limited the described embodiments," which should read "not limited to the described embodiments.".
Appropriate correction is required.
Claim Objections
Claim 2 is objected to because of the following informalities: the claim recites "operation result data" in the first clause and "the operational result data" in the second clause. A single consistent term should be adopted. Appropriate correction is required.
Claim 3 is objected to because of the following informalities: the claim recites "the operational result data," which is inconsistent with "operation result data" as introduced in claim 2, from which claim 3 depends. A single consistent term should be adopted. Appropriate correction is required.
Claim 16 is objected to because of the following informalities: the claim recites "generating the summary data based on the fifth data based on the plurality of similarity values." The doubled "based on" is grammatically defective and apparently should read "generating the summary data from the fifth data based on the plurality of similarity values." Appropriate correction is required.
Claim 17 is objected to because of the following informalities: the claim recites "among the sentences," which is incomplete and apparently should read "from among the sentences of the fifth data." Appropriate correction is required.
Claim 19 is objected to because of the following informalities: the claim recites "the a similarity value," which contains a duplicated article and should read "a similarity value" or "the similarity value." Claim 19 further recites "unsupervised training," which is inconsistent with "unsupervised learning" as recited in claims 1 and 20. Appropriate correction is required.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claim 15-19 rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claim 15 recites the limitation "the failure data." There is insufficient antecedent basis for this limitation in the claim. The failure data of claim 12 is output from the first data in a first inference pass; claim 15's fourth data is output from different input (the third data) after the weight update, so it cannot comprise the same failure data. It is unclear whether identity of content or merely the same type of data is required. Claims 16–19 are rejected by virtue of their dependency from claim 15.
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.
Claim 1-20 rejected under 35 U.S.C. 101 because because the claimed invention is directed to an abstract idea without significantly more.
MPEP 2106 (III) sets out steps for evaluating whether a claim is drawn to patent-eligible subject matter. The analysis of claims 1-20, in accordance with these steps, follows.
Step 1 Analysis:
Claims 1-20 are directed to method (processes). Therefore, claims 1-20 fall into one of four statutory categories (i.e., process).
As to claim 1,
Step 2A Prong 1: this claim recites the following abstract ideas:
performing preprocessing on the first data to generate second data; (the limitation describes organizing and cleaning the collected data into a prepared form, which is a mental process implemented using a pen and paper.)
generating a first token from the second data using a mapping table; (the limitation describes converting the data into tokens by looking up corresponding entries in a mapping table, which is a mental process implemented using a pen and paper.)
generating ... by performing unsupervised learning on ... using the first token. (the limitation describes analyzing the tokens to derive patterns and relationships from the data, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
obtaining first data comprising semiconductor equipment data; (this limitation describes data collection/receiving, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
a semiconductor language model; (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
a language model; (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 12,
Step 2A Prong 1: this claim recites the following abstract ideas:
... obtaining second data comprising failure data ...; (the limitation describes determining failure information from the semiconductor equipment data, which is an evaluation and judgment activity that can be performed as a mental process in the human mind.)
updating a weight of ... based on the first data, the second data, and the correction data. (the limitation describes recalculating a stored weight value based on the data and the corrective feedback, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
inputting first data comprising semiconductor equipment data into an artificial neural network, and ... output from the artificial neural network; (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
obtaining correction data indicating an appropriateness of the second data; (this limitation describes data collection/receiving, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
the artificial neural network; (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 20,
Step 2A Prong 1: this claim recites the following abstract ideas:
performing unsupervised learning on ... using semiconductor equipment data ...; (the limitation describes analyzing the semiconductor equipment data to derive patterns and relationships from the data, which is a mental process implemented using a pen and paper.)
performing supervised learning on ... based on the semiconductor equipment data, label data, and correction data ...; (the limitation describes analyzing the data together with its corresponding labels and corrective feedback to work out relationships between them, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
an artificial neural network; (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
to generate a semiconductor language model; (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
the semiconductor language model; (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
to generate a semiconductor inference model. (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 2,
Step 2A Prong 1: this claim recites the following abstract ideas:
performing the preprocessing on the operational result data to generate the second data. (the limitation describes the content of the data being preprocessed, which merely specifies the type of data evaluated and is an evaluation and judgment activity that can be performed as a mental process in the human mind.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
obtaining operation result data from semiconductor equipment; (this limitation describes data collection/receiving, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 3,
Step 2A Prong 1: this claim recites the following abstract ideas:
performing the preprocessing on the operational result data and the operational reporting data to generate the second data. (the limitation describes the content of the data being preprocessed, which merely specifies the type of data evaluated and is an evaluation and judgment activity that can be performed as a mental process in the human mind.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
obtaining operational reporting data created by a user operating the semiconductor equipment; (this limitation describes data collection/receiving, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 4,
Step 2A Prong 1: this claim recites the following abstract ideas:
generating ... by performing supervised learning on ... based on the third data and the first label data. (the limitation describes analyzing the data together with its corresponding labels to work out relationships between them, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
obtaining third data comprising the semiconductor equipment data, and first label data corresponding to the third data; (this limitation describes data collection/receiving, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
a semiconductor inference model; (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
the semiconductor language model; (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 5,
Step 2A Prong 1: this claim recites the following abstract ideas:
obtaining failure action data for resolving a failure of semiconductor equipment from the third data; (the limitation describes identifying failure-resolution information within the data, which is an evaluation and judgment activity that can be performed as a mental process in the human mind.)
determining a value of the first label data corresponding to the failure action data as a first value. (the limitation describes assigning a label value to the identified data, which is a mental process implemented in the human mind.)
Step 2A Prong 2 and 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B. Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) 1.), failing step 2A prong 2. The claim is ineligible.
As to claim 6,
Step 2A Prong 1: this claim recites the following abstract ideas:
obtaining failure cause data indicating a cause of a failure of semiconductor equipment in the third data; (the limitation describes identifying failure-cause information within the data, which is an evaluation and judgment activity that can be performed as a mental process in the human mind.)
determining a value of the first label data corresponding to the failure cause data as a first value. (the limitation describes assigning a label value to the identified data, which is a mental process implemented in the human mind.)
Step 2A Prong 2 and 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B. Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) 1.), failing step 2A prong 2. The claim is ineligible.
As to claim 7,
Step 2A Prong 1: this claim recites the following abstract ideas:
... obtaining inference data ... based on the third data and the first label data; (the limitation describes drawing an inference from the data and its corresponding labels, which is an evaluation and judgment activity that can be performed as a mental process in the human mind.)
updating a weight of ... based on the third data, the first label data, the inference data, and the correction data. (the limitation describes recalculating a stored weight value based on the data and the corrective feedback, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
inputting the third data and the first label data into the semiconductor inference model, and ... output from the semiconductor inference model; (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
obtaining correction data indicating an appropriateness of the inference data; (this limitation describes data collection/receiving, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
the semiconductor inference model; (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 8,
Step 2A Prong 1: this claim recites the following abstract ideas:
... obtaining inference data ... based on the fourth data and the second label data; (the limitation describes drawing an inference from the data and its corresponding labels, which is an evaluation and judgment activity that can be performed as a mental process in the human mind.)
determining a plurality of similarity values indicating similarities between sentences of the inference data based on a semiconductor terminology database; (the limitation describes calculating similarity values between sentences by consulting a terminology reference, which is a mental process implemented using a pen and paper.)
generating summary data based on the inference data and the plurality of similarity values. (the limitation describes composing a summary of the data based on the calculated similarity values, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
inputting fourth data comprising the semiconductor equipment data, and second label data corresponding to the fourth data, into the semiconductor inference model, and ... output from the semiconductor inference model; (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 9,
Step 2A Prong 1: this claim recites the following abstract ideas:
... a sentence having a largest difference between an importance value and a similarity value from among the sentences in the inference data. (the limitation describes selecting a sentence by comparing importance values and similarity values and identifying the largest difference, which is a mental process implemented in the human mind.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
outputting; (this limitation describes data transmission, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 10,
Step 2A Prong 1: this claim recites the following abstract ideas:
obtaining a first vector corresponding to a first sentence from among the sentences of the inference data, and a second vector corresponding to a second sentence from among the sentences of the inference data; (the limitation describes converting sentences into vector representations, which is a mental process implemented using a pen and paper.)
determining a similarity value from among the plurality of similarity values based on a cosine value of the first vector and a cosine value of the second vector. (the limitation describes calculating cosine values of the vectors and determining a similarity value from them, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B. Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) 1.), failing step 2A prong 2. The claim is ineligible.
As to claim 11,
Step 2A Prong 1: This claim does not recite an additional abstract idea, but the claim depends on claim 10.
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
inputting the first sentence and the second sentence into the semiconductor language model, and obtaining the first vector and the second vector from the semiconductor language model. (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 13,
Step 2A Prong 1: This claim does not recite an additional abstract idea, but the claim depends on claim 12.
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
displaying the second data to a user operating semiconductor equipment; (this limitation describes data transmission, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
receiving the correction data corresponding to the second data from the user. (this limitation describes data collection/receiving, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 14,
Step 2A Prong 1: this claim recites the following abstract ideas:
obtaining inverted data obtained by inverting a value of the second data based on the correction data; (the limitation describes inverting a data value based on the corrective feedback, which is a mental process implemented using a pen and paper.)
updating the weight such that ... outputs the inverted data based on the first data. (the limitation describes recalculating the weight value so that the inverted result is produced from the data, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
the artificial neural network; (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 15,
Step 2A Prong 1: this claim recites the following abstract ideas:
... obtaining fourth data comprising the failure data ...; (the limitation describes determining failure information from the semiconductor equipment data, which is an evaluation and judgment activity that can be performed as a mental process in the human mind.)
generating summary data of the third data based on the fourth data. (the limitation describes composing a summary of the data based on the failure information, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
inputting third data comprising the semiconductor equipment data into the artificial neural network, and ... output from the artificial neural network; (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 16,
Step 2A Prong 1: this claim recites the following abstract ideas:
adding hardware information to the fourth data based on a semiconductor terminology database to generate fifth data; (The recited step represents retrieving hardware information from a terminology reference and adding it to the data, which falls within the abstract idea category of collecting, analyzing, and presenting information (see MPEP 2106.04(a)(2)(iii)).)
determining a plurality of similarity values indicating similarities between sentences of the fifth data; (the limitation describes calculating similarity values between sentences, which is a mental process implemented using a pen and paper.)
generating the summary data based on the fifth data based on the plurality of similarity values. (the limitation describes composing the summary from the data based on the calculated similarity values, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B. Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) 1.), failing step 2A prong 2. The claim is ineligible.
As to claim 17,
Step 2A Prong 1: this claim recites the following abstract ideas:
determining a plurality of importance values corresponding to the sentences of the fifth data ...; (the limitation describes calculating importance values for the sentences, which is a mental process implemented using a pen and paper.)
... a first sentence having a highest importance value from among the sentences of the fifth data; (the limitation describes selecting the sentence with the highest value by comparing the calculated values, which is a mental process implemented in the human mind.)
... a second sentence having a lowest similarity to the first sentence among the sentences. (the limitation describes selecting the sentence with the lowest similarity by comparing the calculated values, which is a mental process implemented in the human mind.)
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
using the artificial neural network; (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
outputting; (this limitation describes data transmission, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
As to claim 18,
Step 2A Prong 1: this claim recites the following abstract ideas:
obtaining a vector corresponding to the fifth data; (the limitation describes converting the data into a vector representation, which is a mental process implemented using a pen and paper.)
determining a similarity value from among the plurality of similarity values based on the vector. (the limitation describes calculating a similarity value from the vector, which is a mental process implemented using a pen and paper.)
Step 2A Prong 2 and 2B:
The claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception under step 2B. Thus, the judicial exception is not integrated into a practical application (see MPEP 2106.04(d) 1.), failing step 2A prong 2. The claim is ineligible.
As to claim 19,
Step 2A Prong 1: This claim does not recite an additional abstract idea, but the claim depends on claim 18.
Step 2A Prong 2 and 2B: the claim recited the following additional elements:
obtaining a semiconductor language model that is trained using unsupervised training based on the semiconductor equipment data to generate a vector corresponding to the semiconductor equipment data; (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
inputting the fifth data into the semiconductor language model to obtain the vector corresponding to the fifth data. (This limitation is directed to mere instruction to apply the abstract idea on a generic computer to process, which is a well-understood, routine, conventional activity, see MPEP 2106.05(d)(II)(i))
The additional element does not integrate the judicial exception into practical application and does not amount to significantly more than the Judicial exception.
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.
Claim(s) 1-6 is/are rejected under 35 U.S.C. 103 as being unpatentable over Ni et al. (US 20240126634 A1) in view of Wu (US 20210383061 A1).
As to claim 1, Ni teaches a language model training method comprising:
performing preprocessing on the first data to generate second data; (see Ni paragraph [0055] "When processing the SR, the title and description thereof are concatenated into a document as an input. All words in the title and description are converted into lower case, and punctuation marks are removed.")
generating a first token from the second data using a mapping table; and (see Ni paragraph [0063] "In the BERT, an input is a token sequence, which is first embedded into a vector and then processed in the stacked encoders.", and see Ni paragraph [0064] "an output vector is multiplied by an embedding matrix and converted into a vocabulary dimension, and the probability of each word in the vocabulary is calculated utilizing softmax.")
generating a semiconductor language model by performing unsupervised learning on a language model using the first token. (see Ni paragraph [0041] "Firstly, at 303, SR training data set SR.sub.TRAIN 301 input to a Masked Language Model (MLM), and the MLM is pre-trained (the MLM will be described in detail later with reference to FIG. 5) to obtain SR-based BERT model BERT.sub.SR 305. ... In this way, a language model represented by an SR is learned.", and see Ni paragraph [0064] "In the stage of pre-training a BERT model, the Masked Language Model is trained together with a Next Sentence Prediction (NSP) model, aiming at minimizing a combined loss function of the two strategies. In the Masked Language Model, before inputting a word sequence to the BERT, 15% of the words in each sequence are replaced with [MASK] tags. The model then attempts to predict an original value of a masked word based on the context provided by other unmasked words in the sequence.")
Ni doesn’t explicitly teach "obtaining first data comprising semiconductor equipment data"
However, Wu teaches
obtaining first data comprising semiconductor equipment data; (see Wu paragraph [0017] "Every time the semiconductor equipment 190 fails, an engineer fixes the semiconductor equipment 190; accordingly, the engineer uses natural language to record a record of failure history of the semiconductor equipment 190 through the computer 170, the semiconductor equipment 190 or another device. Therefore, the records of the failure histories of the semiconductor equipment include natural language.", and see Wu paragraph [0019] "The storage device 110 preloads the records of failure histories of semiconductor equipment 190.")
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify the fault-report language model training of Ni to be performed on Wu's records of failure histories of semiconductor equipment, in order to classify topics of various problems of the semiconductor equipment and provide corresponding solutions, thereby helping the engineer to quickly find the way to fix the semiconductor equipment whenever it fails again, see Wu paragraph [0019].
As to claim 2, Ni as modified by Wu teaches the language model training method of claim 1,
wherein the obtaining the first data comprises obtaining operation result data from semiconductor equipment, and (see Wu paragraph [0018] "For example, the record of the failure history of the semiconductor equipment 190 is described in below. ... 1. When tool run P94911700E ... #23~24 occur tool alarm ... 2. Check alarm message: side1 lamp B UV lamp fault: phase control at limit")
wherein the performing the preprocessing on the first data comprises performing the preprocessing on the operational result data to generate the second data. (see Wu paragraph [0020] "the storage device 110 stores a dictionary file, and the processor 120 is according to the dictionary file to find valid words in the records of the failure histories of the semiconductor equipment 190, so as to generate a record-word matrix that indicates word frequencies of valid words in each of the records of the failure histories of the semiconductor equipment. For example, the dictionary file defines invalid words, and therefore the remaining words in the records are the valid words.")
As to claim 3, Ni as modified by Wu teaches the language model training method of claim 2,
wherein the obtaining the first data further comprises obtaining operational reporting data created by a user operating the semiconductor equipment, and (see Wu paragraph [0017] "Every time the semiconductor equipment 190 fails, an engineer fixes the semiconductor equipment 190; accordingly, the engineer uses natural language to record a record of failure history of the semiconductor equipment 190 through the computer 170")
wherein the performing the preprocessing on the first data further comprises performing the preprocessing on the operational result data and the operational reporting data to generate the second data. (see Wu paragraph [0020] "the processor 120 is according to the dictionary file to find valid words in the records of the failure histories of the semiconductor equipment 190 ... the dictionary file defines invalid words, and therefore the remaining words in the records are the valid words.")
As to claim 4, Ni as modified by Wu teaches the language model training method of claim 1,
further comprising: obtaining third data comprising the semiconductor equipment data, and first label data corresponding to the third data; and (see Ni paragraph [0042] "SR-based BERT model BERT.sub.SR 305 is used to perform multi-class classification and fine-tuning on an SR training set labeled with a knowledge base (KB), that is, LSR (SR labeled with KB) training data set LSR.sub.TRAIN 313")
generating a semiconductor inference model by performing supervised learning on the semiconductor language model based on the third data and the first label data. (see Ni paragraph [0042] "so as to obtain original BTM 310. ... In this way, a classifier recommended for the knowledge base is learned.", and see Ni paragraph [0057] "During the process of fine-tuning a model, a classification layer is added before a Transformer output of a [CLS] token, and an SR-based BERT model (BERT.sub.SR, such as BERT.sub.SR 305 in FIG. 3A) is fine-tuned on an LSR training data set (LSR.sub.TRAIN, such as LSR.sub.TRAIN 313 in FIG. 3A).")
As to claim 5, Ni as modified by Wu teaches the language model training method of claim 4,
wherein the obtaining of the third data and the first label data comprises:
obtaining failure action data for resolving a failure of semiconductor equipment from the third data; and (see Wu paragraph [0018] "3. Reset and standby lamp ok 4. Re-curing #25~01 OK 5. Pass TA dummy OK and tool keep run", and see Wu paragraph [0019] "the abnormal model classification table classifies topics of various problems of the semiconductor equipment 190 and provides corresponding solutions.")
determining a value of the first label data corresponding to the failure action data as a first value. (see Ni paragraph [0054] "When a SR is located and solved, after an appropriate knowledge base (KB) is attached to the SR, for example, an index to the KB being attached to the SR, that is, the SR includes an index to the KB. The KB stores knowledge references for fixing faults. Not all the SRs have corresponding KBs; if the SR does not have a corresponding KB, "NA" may be used to represent an index to a KB.")
As to claim 6, Ni as modified by Wu teaches the language model training method of claim 4,
wherein the obtaining of the third data and the first label data comprises:
obtaining failure cause data indicating a cause of a failure of semiconductor equipment in the third data; and (see Ni paragraph [0021] "These various faults may be caused by software defects, hardware or network problems, or operation mistakes by customers. These various faults will be reported to a customer support team as service requests (SRs) of customers", and see Ni paragraph [0048] "At 390, it is determined utilizing distilled BTM 365 whether a cause of a fault belongs to environmental factors. If distilled BTM 365 determines that the cause of the fault belongs to environmental factors (such as network disconnection, disk damage, power fault, etc.)")
determining a value of the first label data corresponding to the failure cause data as a first value. (see Ni paragraph [0054] "an index to the KB being attached to the SR, that is, the SR includes an index to the KB. ... if the SR does not have a corresponding KB, "NA" may be used to represent an index to a KB.")
Claim(s) 7 is/are rejected under 35 U.S.C. 103 as being unpatentable over Ni et al. (US 20240126634 A1) in view of Wu (US 20210383061 A1), and further in view of Forman et al. (US 20080103996 A1).
As to claim 7, Ni as modified by Wu teaches the language model training method of claim 4, further comprising:
inputting the third data and the first label data into the semiconductor inference model, and obtaining inference data output from the semiconductor inference model based on the third data and the first label data; (see Ni paragraph [0042] "At 319, based on original BTM 310, LSR evaluation data set LSR.sub.EVAL 317 is evaluated to obtain multi-class accuracy 321.", and see Ni paragraph [0055] "The Top 1 precision refers to a maximum value extracted from a final softmax output in multi-class text classification problems, which corresponds to the confidence value of a text input prediction class.")
Ni doesn’t explicitly teaches "obtaining correction data indicating an appropriateness of the inference data; and", and "updating a weight of the semiconductor inference model based on the third data, the first label data, the inference data, and the correction data"
However, Forman teaches
obtaining correction data indicating an appropriateness of the inference data; and (see Forman paragraph [0025] "As used herein, "confirmation/re-labeling" refers to the process of submitting an existing training sample for labeling so that its previously assigned label is either confirmed or contradicted, e.g., by a domain expert or other person. A request for confirmation/re-labeling can include, e.g.: (i) a mere request to indicate whether the previously assigned classification label is correct; and/or (ii) a request to designate a different label if the previously assigned classification label is incorrect. In response, a "reply classification label" is received.")
updating a weight of the semiconductor inference model based on the third data, the first label data, the inference data, and the correction data. (see Forman paragraph [0018] "a training module 5 sets the classification parameters (e.g., weights) of classifier 3 using a set of training samples 7 and class labels 8 that have been assigned to such training samples 7", and see Forman paragraph [0044] "In any event, the confirmation/re-labeling information is received for the submitted training examples and, in step 18, that information is used to modify the training set 45 and then retrain the classifier 3, e.g., using training module 5 and the revised training set 45.")
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to further modify Ni as modified by Wu with Forman's confirmation/re-labeling of the model's outputs by a domain expert, in order to use that information to modify the training set and retrain the classifier, thereby correcting inaccurate labels and improving the accuracy of the classifier's predictions, see Forman paragraph [0044].
Claim(s) 8-9 is/are rejected under 35 U.S.C. 103 as being unpatentable over Ni et al. (US 20240126634 A1) in view of Wu (US 20210383061 A1), and further in view of Fox et al. (US 20210390127 A1).
As to claim 8, Ni as modified by Wu teaches the language model training method of claim 4,
further comprising: inputting fourth data comprising the semiconductor equipment data, and second label data corresponding to the fourth data, into the semiconductor inference model, and obtaining inference data output from the semiconductor inference model based on the fourth data and the second label data; (see Ni paragraph [0042] "At 319, based on original BTM 310, LSR evaluation data set LSR.sub.EVAL 317 is evaluated to obtain multi-class accuracy 321.")
Ni doesn’t explicitly teaches "determining a plurality of similarity values indicating similarities between sentences of the inference data based on a semiconductor terminology database; and", and "generating summary data based on the inference data and the plurality of similarity values"
However, Fox teaches
determining a plurality of similarity values indicating similarities between sentences of the inference data based on a semiconductor terminology database; and (see Fox paragraph [0168] "The keyword recognition component 138 can construct a query by concatenating the best extracted keywords.", and see Fox paragraph [0170] "The keyword recognition component 138 can use TF-IDF scores for each sentence. Then cosine similarity is used to calculate the similarity between two candidate sentences, and between the query and the candidate sentence; these help to remove the redundancy.")
generating summary data based on the inference data and the plurality of similarity values. (see Fox paragraph [0172] "Accordingly, we get a diverse and relevant list of candidate sentences that we can use for summary generation. Using an optimized value of λ, MMR ranking allows minimizing redundancy and leads to query-relevant summarization.")
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to further modify Ni as modified by Wu with Fox's similarity-based summarization of the processed sentences, in order to select and eliminate sentences so the resulting summary sentences are salient, diverse, and relevant, providing the relevant information in a shorter document, see Fox paragraph [0041].
As to claim 9, Ni as modified by Wu and Fox teaches the language model training method of claim 8,
wherein the generating of the summary data comprises outputting a sentence having a largest difference between an importance value and a similarity value from among the sentences in the inference data. (see Fox paragraph [0170] "The MMR formula is given as equation 1. ... MMR = def Arg max Di∈R/S [λ(Sim1(Di,Q)−(1−λ) max Dj∈S Sim2(Di,Dj))]", and see Fox paragraph [0171] "S=Subset of sentences in R already selected. ... λ=Constant used for diversifying the results, in range [0-1].", and see Fox paragraph [0172] "Since we need a combination of both, we set the value of λ toward the middle of the range of 0 to 1. We use a λ value of 0.5 to optimize both the criterion.")
Claim(s) 10-11 is/are rejected under 35 U.S.C. 103 as being unpatentable over Ni et al. (US 20240126634 A1) in view of Wu (US 20210383061 A1) and Fox et al. (US 20210390127 A1), and further in view of Gong et al. (US 20020138528 A1).
As to claim 10, Ni as modified by Wu and Fox teaches the language model training method of claim 8,
determining a similarity value from among the plurality of similarity values based on a cosine value of the first vector and a cosine value of the second vector. (see Fox paragraph [0170] "Then cosine similarity is used to calculate the similarity between two candidate sentences, and between the query and the candidate sentence; these help to remove the redundancy.")
Ni as modified by Wu and Fox does not explicitly teach "wherein the determining of the plurality of similarity values comprises: obtaining a first vector corresponding to a first sentence from among the sentences of the inference data, and a second vector corresponding to a second sentence from among the sentences of the inference data; and"
However, Gong teaches
wherein the determining of the plurality of similarity values comprises: obtaining a first vector corresponding to a first sentence from among the sentences of the inference data, and a second vector corresponding to a second sentence from among the sentences of the inference data; and (see Gong paragraph [0024] "A system and method of generic text summarization may advantageously decompose an entire document into a plurality of individual sentences; after such decomposition, a weighted term-frequency vector may be created for each sentence in the document")
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to further modify the combination with Gong's per-sentence weighted term-frequency vectors for the similarity computation, in order to cover the major topics of the document as thoroughly as possible within a prescribed length while at the same time minimizing redundancy, see Gong paragraph [0023].
As to claim 11, Ni as modified by Wu, Fox, and Gong teaches the language model training method of claim 10, wherein the obtaining of the first vector and the second vector comprises inputting the first sentence and the second sentence into the semiconductor language model, and obtaining the first vector and the second vector from the semiconductor language model. (see Ni paragraph [0063] "In the BERT, an input is a token sequence, which is first embedded into a vector and then processed in the stacked encoders. An output is a vector sequence, where each vector corresponds to input tokens with the same indexes.")
Claim(s) 12, and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Ni et al. (US 20240126634 A1) in view of Forman et al. (US 20080103996 A1).
As to claim 12, Ni teaches a language model training method comprising:
inputting first data comprising semiconductor equipment data into an artificial neural network, and obtaining second data comprising failure data output from the artificial neural network; (see Ni paragraph [0048] "At 390, it is determined utilizing distilled BTM 365 whether a cause of a fault belongs to environmental factors. If distilled BTM 365 determines that the cause of the fault belongs to environmental factors (such as network disconnection, disk damage, power fault, etc.)", and see Ni paragraph [0059] "If a fault occurs to the storage system product, a title and description of the fault is input to the local pre-diagnosis tool (the distilled BTM) to obtain a knowledge base prediction result.")
Ni does not explicitly teaches "obtaining correction data indicating an appropriateness of the second data; and", and "updating a weight of the artificial neural network based on the first data, the second data, and the correction data"
However, Forman teaches
obtaining correction data indicating an appropriateness of the second data; and (see Forman paragraph [0025] "As used herein, "confirmation/re-labeling" refers to the process of submitting an existing training sample for labeling so that its previously assigned label is either confirmed or contradicted, e.g., by a domain expert or other person. ... In response, a "reply classification label" is received.")
updating a weight of the artificial neural network based on the first data, the second data, and the correction data. (see Forman paragraph [0018] "a training module 5 sets the classification parameters (e.g., weights) of classifier 3 using a set of training samples 7 and class labels 8 that have been assigned to such training samples 7", and see Forman paragraph [0044] "In any event, the confirmation/re-labeling information is received for the submitted training examples and, in step 18, that information is used to modify the training set 45 and then retrain the classifier 3, e.g., using training module 5 and the revised training set 45.")
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify Ni's neural network fault diagnosis with Forman's confirmation/re-labeling of the network's outputs, in order to use that information to modify the training set and retrain the classifier, thereby correcting inaccurate labels and improving the accuracy of the classifier's predictions, see Forman paragraph [0044].
As to claim 20, Ni teaches a language model training method comprising:
performing unsupervised learning on an artificial neural network using semiconductor equipment data to generate a semiconductor language model; and (see Ni paragraph [0041] "SR training data set SR.sub.TRAIN 301 input to a Masked Language Model (MLM), and the MLM is pre-trained ... to obtain SR-based BERT model BERT.sub.SR 305.", and see Ni paragraph [0064] "In the stage of pre-training a BERT model, the Masked Language Model is trained together with a Next Sentence Prediction (NSP) model, aiming at minimizing a combined loss function of the two strategies.")
performing supervised learning on the semiconductor language model based on the semiconductor equipment data, label data, and (see Ni paragraph [0042] "SR-based BERT model BERT.sub.SR 305 is used to perform multi-class classification and fine-tuning on an SR training set labeled with a knowledge base (KB), that is, LSR (SR labeled with KB) training data set LSR.sub.TRAIN 313 so as to obtain original BTM 310.", and see Ni paragraph [0057] "During the process of fine-tuning a model, a classification layer is added before a Transformer output of a [CLS] token, and an SR-based BERT model (BERT.sub.SR, such as BERT.sub.SR 305 in FIG. 3A) is fine-tuned on an LSR training data set (LSR.sub.TRAIN, such as LSR.sub.TRAIN 313 in FIG. 3A).")
Ni doesn’t explicitly teach "correction data to generate a semiconductor inference model"
However, Forman teaches
correction data to generate a semiconductor inference model. (see Forman paragraph [0044] "In any event, the confirmation/re-labeling information is received for the submitted training examples and, in step 18, that information is used to modify the training set 45 and then retrain the classifier 3, e.g., using training module 5 and the revised training set 45.")
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to modify Ni's supervised fine-tuning with Forman's incorporation of confirmation/re-labeling information into the training set, in order to modify the training set and retrain the classifier, thereby correcting inaccurate labels and improving the accuracy of the classifier's predictions, see Forman paragraph [0044].
Claim(s) 13-14 is/are rejected under 35 U.S.C. 103 as being unpatentable over Ni et al. (US 20240126634 A1) in view of Forman et al. (US 20080103996 A1), and further in view of Wu (US 20210383061 A1).
As to claim 13, Ni as modified by Forman teaches the language model training method of claim 12,
receiving the correction data corresponding to the second data from the user. (see Forman paragraph [0025] "In response, a "reply classification label" is received. Such a reply classification label can comprise a mere designation (explicit or implicit) that the previously assigned classification label is correct and/or a different classification label which is believed to be more appropriate than the previously assigned classification label.")
Ni as modified by Forman does not explicitly teach "wherein the obtaining the correction data comprises: displaying the second data to a user operating semiconductor equipment; and"
However, Wu teaches
wherein the obtaining the correction data comprises: displaying the second data to a user operating semiconductor equipment; and (see Wu paragraph [0019] "the communication device 130 transmits the abnormal model classification table to the computer 170, so that the display screen 172 of the computer 170 can display the abnormal model classification table, where the abnormal model classification table classifies topics of various problems of the semiconductor equipment 190 and provides corresponding solutions. Thus, the abnormal model classification table can help the engineer to quickly find the way to fix the semiconductor equipment 190")
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to further modify Ni as modified by Forman by displaying the model's classification output to the engineer as taught by Wu, in order to help the engineer to quickly find the way to fix the semiconductor equipment whenever the semiconductor equipment fails again, see Wu paragraph [0019].
As to claim 14, Ni as modified by Forman and Wu teaches the language model training method of claim 13,
wherein the updating the weight of the artificial neural network comprises: obtaining inverted data obtained by inverting a value of the second data based on the correction data; and (see Forman paragraph [0046] "the accuracy presumption 66 preferably is much higher than accuracy presumption 65, with the result that the new label 88 assigned by user 57 replaces the previous label 8.", and see Forman paragraph [0058] "The measure of the effect can be determined by assuming the classification of each sample would be reversed (or, for non-binary classification problems, changed to the next closest category) if submitted for labeling 180")
updating the weight such that the artificial neural network outputs the inverted data based on the first data. (see Forman paragraph [0018] "a training module 5 sets the classification parameters (e.g., weights) of classifier 3", and see Forman paragraph [0047] "In any event, once the training set 45 has been updated, the classifier 3 is retrained using the modified training set 45 (e.g., by training module 5), and the retrained classifier 3 is used to re-process at least some of the labeled training samples 45, thereby obtaining new predictions 4.")
Claim(s) 15-16, and 18-19 is/are rejected under 35 U.S.C. 103 as being unpatentable over Ni et al. (US 20240126634 A1) in view of Forman et al. (US 20080103996 A1), and further in view of Fox et al. (US 20210390127 A1).
As to claim 15, Ni as modified by Forman teaches the language model training method of claim 12,
further comprising: inputting third data comprising the semiconductor equipment data into the artificial neural network, and obtaining fourth data comprising the failure data output from the artificial neural network; and (see Ni paragraph [0059] "If a fault occurs to the storage system product, a title and description of the fault is input to the local pre-diagnosis tool (the distilled BTM) to obtain a knowledge base prediction result.")
Ni as modified by Forman does not explicitly teaches "generating summary data of the third data based on the fourth data"
However, Fox teaches
generating summary data of the third data based on the fourth data. (see Fox paragraph [0041] "Frequency and probabilistic techniques can be used to score and/or rank the sentences further, e.g., based on the prevalence of entities and keywords contained in them. The top-n % of sentences from this ranked list can be used as a list of candidate summary sentences. This list can be further trimmed using, possibly iterative, techniques like Maximal Marginal Relevance (MMR) to select and/or eliminate sentences, so the resulting summary sentences are salient, diverse, and relevant.")
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to further modify Ni as modified by Forman with Fox's summary generation from the model-scored sentences, in order to select and eliminate sentences so the resulting summary sentences are salient, diverse, and relevant, providing the relevant information in a shorter document, see Fox paragraph [0041].
As to claim 16, Ni as modified by Forman and fox teaches the language model training method of claim 15,
wherein the generating the summary data comprises: adding hardware information to the fourth data based on a semiconductor terminology database to generate fifth data; (see Fox paragraph [0031] "The file generated in the previous step is processed by the aspect classification component, which generates another output file, that adds to each row an additional column for the aspect of the QA pair.", and see Fox paragraph [0113] "In order to add domain adaptation to summarization, aspect categories can be defined for the domain, field, or topic of a document... The aspects or aspect categories can help create a richer representation for a sentence since they can represent a sentence based on the aspects it contains.", and see Fox paragraph [0115] "A dictionary of words can be created that associates each with a single aspect, as determined below.", and see Fox paragraph [0117] "For each sentence, the aspect for each word can be retrieved from the dictionary created from the training set. The value of the a.sup.th aspect can be incremented by 1 each time a word associated with that aspect was encountered. As shown in the table, vector representations of the sentences can be formed based on the aspect instance counts.")
determining a plurality of similarity values indicating similarities between sentences of the fifth data; and (see Fox paragraph [0116] "To calculate the coherence score, the cosine similarity between two |A|-dimensional vectors can be identified.")
generating the summary data based on the fifth data based on the plurality of similarity values. (see Fox paragraph [0172] "Accordingly, we get a diverse and relevant list of candidate sentences that we can use for summary generation. Using an optimized value of λ, MMR ranking allows minimizing redundancy and leads to query-relevant summarization")
As to claim 18, Ni as modified by Forman and fox teaches the language model training method of claim 16, wherein the determining of the plurality of similarity values comprises:
obtaining a vector corresponding to the fifth data; and (see Fox paragraph [0114] "A sentence can be represented as a combination of aspects, as a vector A, where |A| is the number of aspects.")
determining a similarity value from among the plurality of similarity values based on the vector. (see Fox paragraph [0116] "To calculate the coherence score, the cosine similarity between two |A|-dimensional vectors can be identified.")
As to claim 19, Ni as modified by Forman and fox teaches the language model training method of claim 18, wherein the determining of the plurality of similarity values comprises:
obtaining a semiconductor language model that is trained using unsupervised training based on the semiconductor equipment data to generate a vector corresponding to the semiconductor equipment data; (see Ni paragraph [0041] "SR training data set SR.sub.TRAIN 301 input to a Masked Language Model (MLM), and the MLM is pre-trained ... to obtain SR-based BERT model BERT.sub.SR 305. ... In this way, a language model represented by an SR is learned.", and see Ni paragraph [0063] "An output is a vector sequence, where each vector corresponds to input tokens with the same indexes.")
inputting the fifth data into the semiconductor language model to obtain the vector corresponding to the fifth data; and (see Ni paragraph [0063] "In the BERT, an input is a token sequence, which is first embedded into a vector and then processed in the stacked encoders. An output is a vector sequence, where each vector corresponds to input tokens with the same indexes.")
determining the a similarity value from among the plurality of similarity values based on the vector. (see Fox paragraph [0170] "Then cosine similarity is used to calculate the similarity between two candidate sentences, and between the query and the candidate sentence; these help to remove the redundancy.")
Claim(s) 17 is/are rejected under 35 U.S.C. 103 as being unpatentable over Ni et al. (US 20240126634 A1) in view of Forman et al. (US 20080103996 A1) and Fox et al. (US 20210390127 A1), and further in view of Gong et al. (US 20020138528 A1).
As to claim 17, Ni as modified by Forman and fox teaches the language model training method of claim 16, wherein the generating the summary data comprises:
determining a plurality of importance values corresponding to the sentences of the fifth data using the artificial neural network; (see Ni paragraph [0055] "The Top 1 precision refers to a maximum value extracted from a final softmax output in multi-class text classification problems, which corresponds to the confidence value of a text input prediction class.")
outputting a second sentence having a lowest similarity to the first sentence among the sentences. (see Fox paragraph [0172] "As per the MMR criterion, the algorithm incrementally retrieves the standard-relevance list (sentences more similar to query) when λ=0 or a maximal diversity list (sentences more diverse or dissimilar to other candidate summary sentences) when λ=1.")
Ni as modified by Forman and fox doesn’t explicitly teach "outputting a first sentence having a highest importance value from among the sentences of the fifth data; and"
However, Gong teaches
outputting a first sentence having a highest importance value from among the sentences of the fifth data; and (see Gong paragraph [0039] "Next, a relevance score may be computed for each sentence in the candidate sentence set in accordance with relevance to the document as a whole, and the sentence with the highest relevance score may be selected as a sentence for inclusion in the summary (blocks 103 and 104).")
It would have been obvious to a person of ordinary skill in the art before the effective filing date of the invention to further modify the combination with Gong's selection of the highest-relevance sentence first, in order to ensure that the summary represents the major topics of the document to the greatest extent possible while retrieval of the subsequent sentence creates minimum overlap, achieving a very low level of redundancy, see Gong paragraph [0043].
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ABDULLAH K ABOUD whose telephone number is (571)272-0025. The examiner can normally be reached Mon-Fri 8am-5pm.
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, Li B Zhen, can be reached at (571) 272-3768. 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.
/ABDULLAH KHALED ABOUD/ Examiner, Art Unit 2121
/Li B. Zhen/ Supervisory Patent Examiner, Art Unit 2121
