[AltContent: rect] DETAILED ACTION
Claims 1-25 are presented for examination
This office action is in response to submission of application on 9-MARCH-2022.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Amendment
The amendment filed on 10-FEBURARY-2026 in response to the final office action mailed 13-NOVEMBER-2025 has been entered. Claims 1-25 remain pending in the application.
With regards to the 101 rejection, the rejection to claim 1 has not been overcome by the applicant’s amendments. Despite applicant’s amendments, claim 1 still remains rejected under 35 U.S.C. 101 on the basis of being an abstract idea.
With regards to the 103 rejections, the applicant’s amendments to the claims have not overcome the rejections to claims 1-25 as the former prior art sufficiently teaches the newly added limitations of the amended claims.
Claim Rejections - 35 USC § 101
Claims 1-25 are rejected under 35 U.S.C. 101 because the claimed invention is directed to
an abstract idea (mental process) without significantly more.
Regarding claim 1, in Step 1 of the 101 analysis set forth in MPEP 2106, the claim recites a method that preforms knowledge graph embedding. A device is one of the four statutory categories of
invention.
In Step 2a Prong 1 of the 101 analysis set forth in the MPEP 2106, the examiner has determined
that the following limitations recite a process that, under the broadest reasonable interpretation, covers
a mental process but for recitation of generic computer components:
selecting a node pair for each triple of a knowledge graph (a person can mentally use the data pair to make a selection as a process of simply evaluating the data and making a judgement on the data pair.)
identifying a direct relation path between said node pair, wherein said triple of said knowledge graph comprises a first and a second node each representing an entity connected by a specific relation path; (a person can mentally use the specific relation paths to identify the direct one as a process of simply evaluating the data and making a judgement on the relation paths.)
counting a number of occurrences of each relation path for each triple in said collected set of relation paths thereby forming a feature vector set for each triple, wherein said feature vector set comprises a set of occurrences of each relation path for a node pair along with a corresponding direct relation path; (a person can mentally use the data of the triples to count the reappearing triples as a process of simply evaluating the data and making a judgement on the repeating data.)
predicting a direct relation path between two target nodes in said knowledge graph using said prediction model by obtaining a feature vector set corresponding to said two target nodes. (a person can mentally use the data given to predict a relation path between two nodes as a process of simply evaluating the data and making a judgement on the data.)
If claim limitations, under their broadest reasonable interpretation, covers performance of the
limitations as a mental process but for the recitation of generic computer components, then it falls
within the mental process grouping of abstract ideas. According, the claim “recites” an abstract idea.
In Step 2a Prong 2 of the 101 analysis set forth in MPEP 2106, the examiner has determined that
the following additional elements do not integrate this judicial exception into a practical application:
A computer-implemented method for knowledge graph embedding, the method comprising: (Generally linking the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h))
collecting a set of relation paths between said node pair except for a path representing said direct relation path for each triple of said knowledge graph; (Adding insignificant extra-solution activity (mere data gathering) to the judicial exception (MPEP 2106.05(g))
and constructing a prediction model by using said feature vector set for each triple (In step 2A prong 2, constructing a model is a mere application of a computer tool (M.L. model), which is not indicative of integration into a practical application. In step 2B, merely applying a computer tool is not indicative of significantly more.)
Since the claim does not contain any other additional elements that are indicative of integration
into a practical application, the claim is “directed” to an abstract idea.
In Step 2b of the 101 analysis set forth in the 2019 PEG, the examiner has determined that the
claim does not include additional elements that are sufficient to amount to significantly more than the judicial exception.
As discussed above, the additional elements (iv) recites generally linking the use of the judicial
exception to a particular technological environment or field of use, (v) recites adding insignificant extra-solution activity, and (vi) recites a mere application of a computer tool, which is not indicative of significantly more. Considering the additional elements individually and in combination, and the claim as a whole, the additional elements do not provide significantly more than the abstract idea. Therefore, the claim is not patent eligible.
Regarding claim 2, it is dependent upon claim 1, and thereby incorporates the limitations of, and
corresponding analysis applied to claim 1. Further, claim 2 recites The method as recited in claim 1, wherein each relation path of said set of relation paths forms a closed loop in said knowledge graph. ((Generally linking the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)) Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Regarding claim 3, it is dependent upon claim 1, and thereby incorporates the limitations of, and
corresponding analysis applied to claim 1. Further, claim 3 recites The method as recited in claim 1, wherein each relation path of said set of relation paths does not include node information. (Generally linking the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)) Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Regarding claim 4, it is dependent upon claim 1, and thereby incorporates the limitations of, and
corresponding analysis applied to claim 1. Further, claim 4 recites The method as recited in claim 1 further comprising: obtaining a string representing a relation path. (Adding insignificant extra-solution activity (mere data gathering) to the judicial exception (MPEP 2106.05(g)) Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Regarding claim 5, it is dependent upon claim 4, and thereby incorporates the limitations of, and
corresponding analysis applied to claim 4. Further, claim 5 recites The method as recited in claim 4 further comprising: calculating hash values for said string. (In step 2A, prong 1, this recites a mathematical concept but for recitation of generic computer components which is not indicative of integration into a practical application.) Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Regarding claim 6, it is dependent upon claim 5, and thereby incorporates the limitations of, and
corresponding analysis applied to claim 5. Further, claim 6 recites The method as recited in claim 5, wherein said hash values are calculated using a plurality of hash algorithms. (Generally linking the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)) Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Regarding claim 7, it is dependent upon claim 5, and thereby incorporates the limitations of, and
corresponding analysis applied to claim 5. Further, claim 7 recites The method as recited in claim 5 further comprising: 2 counting a number of occurrences of a same hash value for each of said calculated hash 3 values. (In step 2A, prong 1, this recites a mental process but for recitation of generic computer components which is not indicative of integration into a practical application.) Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Regarding claim 8, it is dependent upon claim 7, and thereby incorporates the limitations of, and
corresponding analysis applied to claim 7. Further, claim 8 recites The method as recited in claim 7 further comprising; selecting a minimum number of occurrences of said same hash value to be used to identify a number of occurrences of a corresponding relation path in said feature vector set for a triple. (In step 2A, prong 1, this recites a mental process but for recitation of generic computer components which is not indicative of integration into a practical application.) Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Regarding claim 9, it is dependent upon claim 1, and thereby incorporates the limitations of, and
corresponding analysis applied to claim 1. Further, claim 9 recites The method as recited in claim 1, wherein said prediction model corresponds to a decision tree. (Generally linking the use of the judicial exception to a particular technological environment or field of use (MPEP 2106.05(h)) Since the claim does not recite additional elements that either integrate the judicial exception into a practical application, nor provide significantly more than the judicial exception, the claim is not patent eligible.
Regarding claims 10-25, they comprise of limitations similar to those of claims 1-9 and are therefore rejected for similar rationale.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
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) 1-3, 10-12, and 19-21 are rejected under 35 U.S.C. 103 as being unpatentable over MINERVINI (U.S. Pat. No. US 20180144252 A1) in view of KASNECI (U.S. Pat. No. US 20120158791 A1) in view of CREED (U.S. Pat. No. 20210081717 A1).
Regarding claim 1, MINERVINI teaches the invention as substantially claimed, including:
A computer-implemented method for knowledge graph embedding, the method comprising: selecting a node pair for each triple of a knowledge graph and identifying a direct relation path between said node pair, wherein said triple of said knowledge graph comprises a first and a second node each representing an entity connected by a specific relation path; ([0020] In the application of pre-processing to the statements forming a knowledge graph, the statements are updated prior to assigning any embedding vectors for the knowledge graph. As an example of this, in the case of “equivalent” relations, for all the triples matching the pattern <s, Partner Of, o>, a triple <s, Married To, o> could be added to the knowledge graph. Similarly, in the case of “inverse” relations, for all the triples matching the pattern <s, Has Part, o>, a triple <o, Part Of, s> could be added to the knowledge graph.) collecting a set of relation paths between said node pair except for a path representing said direct relation path for each triple of said knowledge graph; ([0027] An aspect of an embodiment of the present invention provides a method for completing a knowledge graph from a plurality of predicates and associated entities, the predicates each providing information on a relationship between a pair of entities, the method comprising the steps of:receiving an input comprising the plurality of predicates(i.e. direct paths) and associated entities(i.e. Node pairs); searching an axiom database and identifying predicates among the plurality of predicates that are equivalent to one another, or inverses of one another; identifying further predicates that are related to one another, using the axiom database and identified predicates; and embedding the identified predicates and associated entities into a vector space to complete the knowledge graph (i.e. finding and applying the relation paths that are not the direct one))
While MINERVINI does teach collecting and identifying paths between two nodes, it does not explicitly teach:
counting a number of occurrences of each relation path for each triple in said collected set of relation paths thereby forming a feature vector set for each triple, wherein said feature vector set comprises a set of occurrences of each relation path for a node pair along with a corresponding direct relation path;
However, in analogous art that similarly teaches a method for knowledge graph embedding, KASNECI teaches:
counting a number of occurrences of each relation path for each triple in said collected set of relation paths thereby forming a feature vector set for each triple, wherein said feature vector set comprises a set of occurrences of each relation path for a node pair along with a corresponding direct relation path; ([0028] The vector processing module 206 is also configured to construct a set including each possible returned sub-graph for the entity "E" of type "T" as a set of sub-graphs for entity E (the entity of interest). In an implementation, the feature vector 108 constructed from this information by the vector processing module 206 is configured as a feature vector 108 that has length equal to a number of the possible sub-graph features available for entity "E" of type "T." The feature vector 108 is formed to include indicator variables to describe observance of a feature represented by the respective indicator variables.)
It would have been obvious to a person skilled in the art before the effective filing date of the
invention to have combined with MINERVINI‘s collects and identifies relation paths and, with KASNECI ‘s teaching of counting the occurrences to make a feature vector to realize, with a reasonable expectation of success, a method that collects and identifies relation path, as in MINERVINI, and counts the number of occurrences in them to make a feature vector, as in KASNECI. A person of ordinary skill would have been motivated to make this combination to make the process easier and more accessible to perform. (KASNECI [0002]).
While MINERVINI, as modified by KASNECI, does teach collecting relation paths and counting the occurrences of them, it does not explicitly teach:
and constructing a prediction model by using said feature vector set for each triple; and predicting a direct relation path between two target nodes in said knowledge graph using said prediction model by obtaining a feature vector set corresponding to said two target nodes.
However, in analogous art that similarly performs knowledge graph embedding, CREED teaches:
and constructing and training a prediction model using a machine learning algorithm to predict an unknown direct relation path between two target nodes in said knowledge graph based on sample data consisting of said feature vector set for each triple; ([0074] The invention relates to a system, apparatus and method(s) for efficiently generating and training a robust graph neural network (GNN) model using attention weights and based on data representative of at least a portion of an entity-entity graph dataset generated from, by way of example only but not limited to, a dataset of facts and/or entity relations/relationships and the like; [0083] Although details of the present disclosure are described, by way of example only but is not limited to, by way of reference to GNN models using GCNN, it is to be appreciated by the skilled person that other GNN and/or NN structures, and/or combinations thereof may be applied without loss of generality to the present invention as the application demands. For example, one or more neural network structures may be used to train a GNN to generate a trained GNN model based on an entity-entity graph dataset. The trained GNN model being for use, by way of example only but not limited to, link prediction problems and the like. NNs may comprise or represent one or more of a combination of neural network hidden layers, algorithms, structures and the like that can be trained to generate a GNN model associated with predicting links or relationships between one or more entities of the entity-entity graph from which the GNN model is generated. [0091] The encoding network (or encoder) may be based on any suitable neural network that is capable of generating one or more embeddings of at least a portion of the entity-entity graph. At the end of the encoding process (output of the encoding network), each entity of at least the portion of the entity-entity graph may be represented, by way of example only but is not limited to, an embedding vector (e.g. an N-dimensional embedding vector). [0124] A model formation of the GCNN 200 is derived based on the model formulation of the conventional GCNN, which is based on computing a link prediction that includes two main steps: first, an encoding step (e.g. encoder network) for calculation of an embedding of entities of a entity-entity knowledge graph [AltContent: rect], and second, a decoding step (e.g. scoring network) to calculate the likelihood of a link or relationship between entities. Both steps, however, are linked as an end-to-end differentiable process. (a vector set is made for each entity-entity relation for embedding which is then decoded to make a prediction. In other words, a vector set of the triples corresponding to the target for prediction is generated through encoding then predicted through decoding) ) and predicting a direct relation path between two target nodes in said knowledge graph using said prediction model. ([0074] the GNN model may be used for prediction tasks such as, by way of example only but not limited to, link prediction or relationship inference [0091] The encoding network (or encoder) may be based on any suitable neural network that is capable of generating one or more embeddings of at least a portion of the entity-entity graph. At the end of the encoding process (output of the encoding network), each entity of at least the portion of the entity-entity graph may be represented, by way of example only but is not limited to, an embedding vector (e.g. an N-dimensional embedding vector). [0124] A model formation of the GCNN 200 is derived based on the model formulation of the conventional GCNN, which is based on computing a link prediction that includes two main steps: first, an encoding step (e.g. encoder network) for calculation of an embedding of entities of a entity-entity knowledge graph [AltContent: rect], and second, a decoding step (e.g. scoring network) to calculate the likelihood of a link or relationship between entities. Both steps, however, are linked as an end-to-end differentiable process. (a vector set is made for each entity-entity relation for embedding which is then decoded to make a prediction. In other words, a vector set of the triples corresponding to the target for prediction is generated through encoding then predicted through decoding)
It would have been obvious to a person skilled in the art before the effective filing date of the
invention to have combined with MINERVINI‘s, as modified by KASNECI, method that collects and identifies relation paths then counts the occurrences of said paths and, with CREED ‘s teaching of link prediction, with a reasonable expectation of success, a method that collects and identifies relation path and counts occurrences to make a feature vector, as in MINERVINI, as modified by KANECI, and uses the feature vector to train a model used for link prediction, as in CREED. A person of ordinary skill would have been motivated to make this combination to decrease the noise and increase the accuracy of the model. (CREED [0005]).
Regarding claim 2, MINERVINI further teaches:
The method as recited in claim 1, wherein each relation path of said set of relation paths forms a closed loop in said knowledge graph. ([0043] The present invention provides an addition to existing schema unaware mechanisms for embedding statements (triples) into knowledge graphs. The following description refers to the use of RDF triples (RDF triples fit the definition of a closed loop as provided by the applicant, therefore all relation paths gathered by MINERVINI’s teachings will be closed loops as they follow this format), however the invention is equally applicable to other models that deviate from the RDF requirements. As such, the present invention can be applied to any information (triple) using the <entity, predicate, entity> form.)
Regarding claim 3, MINERVINI further teaches:
The method as recited in claim 1, wherein each relation path of said set of relation paths does not include node information. (receiving an input comprising the plurality of predicates(i.e. direct paths) and associated entities(i.e. Node pairs); searching an axiom database and identifying predicates among the plurality of predicates that are equivalent to one another, or inverses of one another; identifying further predicates that are related to one another, using the axiom database and identified predicates; (The identified predicates are the set of relation paths found in the database that do not include node information) and embedding the identified predicates and associated entities into a vector space to complete the knowledge graph)
Regarding claims 10-12 and 19-21, they comprise of limitations similar to those of claims 1-3 and are therefore rejected for similar rationale.
Claim(s) 4-6, 13-15, and 22-24 are rejected under 35 U.S.C. 103 as being unpatentable over MINERVINI (U.S. Pub. No. US 20180144252 A1), KASNECI (U.S. Pub. No. US 20120158791 A1), CREED (U.S. Pub. No. US 20210081717 A1). In further view of AASMAN (U.S. Pub. No. US 20080243770 A1)
Regarding claim 4, while MINERVINI, as modified by KASNECI and CREED does teach claim 1, which claim 4 is dependent on, it does not explicitly teach:
The method as recited in claim 1 further comprising: obtaining a string representing a relation path
However, in analogous art that similarly performs knowledge graph embedding, AASMAN teaches:
The method as recited in claim 1 further comprising: obtaining a string representing a relation path. ([0030] In the semantic web context, for the first example triple, "FRANZ is a company," the text string "FRANZ" is the subject, the text string "is a" is the predicate, and the text string "company" is the object.)
It would have been obvious to a person skilled in the art before the effective filing date of the
invention to have combined with MINERVINI‘s, as modified by KASNECI and CREED, method that collects and identifies relation paths then counts the occurrences of said paths and performs link prediction with that data and, with AASMAN ‘s obtaining a string correlating with the path, with a reasonable expectation of success, a method that collects and identifies relation path and counts occurrences to make a feature vector and uses that data for link prediction, as in MINERVINI, as modified by KANECI and CREED, and translates the relation path into a string, as in AASMAN. A person of ordinary skill would have been motivated to make this combination to decrease the noise and increase the efficiency of the relation path storage. (AASMAN [0018]).
Regarding claim 5, AASMAN further teaches:
The method as recited in claim 4 further comprising: calculating hash values for said string. ([0036] For short strings, which can be defined as strings where strlen(string)<sizeof(triplepart)-2, the first byte could be assigned a value of "2", the second byte (byte 2) could be coded with the string length, and bytes 3 through 12 could be encoded with the short string in utf8 (8-bit Unicode Transformation Format). For long strings, the first byte (byte 1) could be assigned a value of "3" and bytes 2 through 12 could be could be an 88 bit CRC or any other hash value that guarantees a satisfactory distribution.)
Regarding claim 6, AASMAN further teaches:
The method as recited in claim 5, wherein said hash values are calculated using a plurality of hash algorithms. ([0038] As described above, the triple-parts can be URI's or parts or specialized datatypes. Thus, block 202 essentially performs a part-to-UPI mapping operation. This process of encoding the triple parts with UPIs, also referred to as "interning," involves the step of checking to determine whether that particular UPI has been recently encountered. In one embodiment, a clock cache, or similar buffer mechanism is used. A clock cache is a FIFO (First In-First Out) buffer structure that stores a pre-defined number of entries. Alternatively, each UPI can be referenced in an in-memory hashtable to see if it was recently encountered. The size of any internal hashtable that is used to determine the presence of any duplicate part-to-UPI mappings can be set by the user, and can be regularly refreshed.)
Regarding claims 13-15 and 22-24, the comprise of limitations similar to those of claims 4-6 and are therefore rejected for similar rationale.
Claim(s) 7-8, 16-17, and 25 are rejected under 35 U.S.C. 103 as being unpatentable over MINERVINI (U.S. Pub. No. US 20180144252 A1), KASNECI (U.S. Pub. No. US 20120158791 A1), CREED (U.S. Pub. No. US 20210081717 A1), and AASMAN (U.S. Pub. No. US 20080243770 A1) In further view of JOHNSON (U.S. Pub. No. US 20210272040 A1)
While MINERVINI, as modified by KASNECI, CREED, and AASMAN, does teach claim 5, which claim 7 is dependent on, it does not explicitly teach:
The method as recited in claim 5 further comprising: counting a number of occurrences of a same hash value for each of said calculated hash values.
However, in analogous art that similarly performs knowledge graph embedding, JOHNSON teaches:
The method as recited in claim 5 further comprising: 2 counting a number of occurrences of a same hash value for each of said calculated hash 3 values.( [0327] Frequency distribution identifier 2618 can tokenize and parse the surviving text into N-grams. The N-grams can be cleaned by frequency distribution identifier 2618 to remove punctuation, stop words and digits at the beginning and end of phrases, etc. Frequency distribution identifier 2618 may apply other N-gram processing operations such as checking word length, term length, etc.
[0328] In some embodiments, frequency distribution identifier(i.e. the function the finds the frequency/number of occurrences of the same hash value) 2618 places N-grams(i.e. hashes) and their value counts(i.e. hash values) in a hash map (we know the identifier is a hash table and the values hash values due to them being input into a hash map). If frequency distribution identifier 2618 places the N-grams and their value counts in the hash map, frequency distribution identifier 2618 may apply a map reduce algorithm to the hash table to merge duplicate N-grams. The map reduce algorithm can be ran in parallel to reduce processing time. An output of the map reduce algorithm may be a dictionary of unique N-grams and value counts.)
It would have been obvious to a person skilled in the art before the effective filing date of the
invention to have combined with MINERVINI‘s, as modified by KASNECI, CREED, and AASMAN, method that collects and identifies relation paths then counts the occurrences of said paths and performs link prediction with that data where the relations have hashes assigned to them and, with JOHNSON ‘s counting of reoccurring hashes, with a reasonable expectation of success, a method that collects and identifies relation path and counts occurrences to make a feature vector and uses that data for link prediction and translates relations to hashes, as in MINERVINI, as modified by KANECI, CREED, and AASMAN, and counts the reoccurring hashes, as in JOHNSON. A person of ordinary skill would have been motivated to make this combination to aid in improving performance. (JOHNSON [0329]).
Regarding claim 8, JOHNSON further teaches:
The method as recited in claim 7 further comprising; selecting a minimum number of occurrences of said same hash value to be used to identify a number of occurrences of a corresponding relation path in said feature vector set for a triple. ([0329] Frequency distribution identifier 2618 can calculate frequency distributions by comparing the dictionary of N-grams to a set of tags provided to content summarization engine 2600.)
Regarding claims 16-17, they comprise of limitation similar to those of claims 7-8 and are therefore rejected for similar rationale. Regarding claim 25, is comprises of limitations similar to those of claim 7 and is therefore rejected for similar rationale.
Claim(s) 9 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over MINERVINI (U.S. Pub. No. US 20180144252 A1), KASNECI (U.S. Pub. No. US 20120158791 A1), CREED (U.S. Pub. No. US 20210081717 A1). In further view of CHICKERING (U.S. Pub. No. US 20050131848 A1)
While MINERVINI, as modified by KANSECI and CREED, does teach claim 1, which claim 9 is dependent on, it does not explicitly teach:
The method as recited in claim 1, wherein said prediction model corresponds to a decision tree.
However, in analogous art that similarly performs knowledge graph embedding, CHICKERING teaches:
The method as recited in claim 1, wherein said prediction model corresponds to a decision tree. ([0026] The present invention provides improved data mining systems and methods of generating a Bayesian network, employing scalable algorithms that learn individual local distributions (e.g., decision trees) to build a Bayesian network (i.e., a set of local distributions that contains no cycle of predictor to target relations) in a scalable manner.)
It would have been obvious to a person skilled in the art before the effective filing date of the
invention to have combined with MINERVINI‘s, as modified by KASNECI and CREED, method that collects and identifies relation paths then counts the occurrences of said paths and performs link prediction with that data and, with CHICKERING ‘s decision tree prediction model, with a reasonable expectation of success, a method that collects and identifies relation path and counts occurrences to make a feature vector and uses that data for link prediction, as in MINERVINI, as modified by KANECI and CREED, where the prediction model uses a decision tree, as in CHICKERING. A person of ordinary skill would have been motivated to make this combination to increase scalability of the model. (CHICKERING [0011]).
Response to Arguments
Applicant’s arguments filed 10-FEBURARY-2026 have been fully considered, but they are found to be non-persuasive
With regards to the applicant’s remarks regarding the 101 rejection towards an abstract idea, the applicant argues that the amendments to claim 1 overcome the rejection of identifying, counting, selecting, and identifying data
as recited in claim 1 and similarly in claims 10 and 19, collecting a set of relation paths between the node pair except for a path representing the direct relation path for each triple of the knowledge graph; counting a number of occurrences of each relation path for each triple in the collected set of relation paths thereby forming a feature vector set for each triple, where the feature vector set comprises a set of occurrences of each relation path for a node pair along with a corresponding direct relation path; constructing and training a prediction model using a machine learning algorithm to predict an unknown direct relation path between two target nodes in the knowledge graph based on sample data consisting of the feature vector set for each triple; and predicting the direct relation path between the two target nodes in the knowledge graph using the prediction model, cannot practically be performed in the human mind, including using a pen and paper. That is, collecting relation paths, forming feature vector sets, constructing/training a prediction model using a machine learning algorithm, and predicting a direct relation path using the prediction model cannot practically be performed in the human mind. Because these claim limitations are not "practically" performable by a human, they do not recite a judicial exception. That is, such claim limitations do not recite the judicial exception (abstract idea) of a mental process.."
With regards to this argument, the examiner acknowledges that these processes are performed by a model. However, simply because these processes are performed by a prediction model does not mean that a human is incapable of performing them in their mind or with a piece of paper and pencil. No reason is given as to why a human would be incapable of performing these actions.
Hence, Furthermore, for each triple of the knowledge graph, a set of relation paths between the selected node pair is collected except for the path representing the direct relation path. The number of occurrences of each relation path for each triple in the collected set of relation paths is counted thereby forming a feature vector set for each triple, where the feature vector set includes a set of occurrences of each relation path for a node pair along with a corresponding direct relation path. A prediction model is then constructed using the feature vector set for each triple to predict an unknown direct relation path between two target nodes in the knowledge graph by obtaining a feature vector set corresponding to the two target nodes, which includes the number of occurrences for various relation paths as well as a direct relation path. Based on obtaining the number of occurrences of each relation path connecting the two target nodes, the unknown direct relation path can be predicted for the two target nodes based on obtaining the feature vector set of the two target nodes containing the number of occurrences (or most similar number of occurrences) of each relation path (or most similar relation path) connecting the two target nodes. In this manner, knowledge graph embedding is performed more accurately and efficiently than prior techniques. The technique of the present disclosure utilizes fewer resources (e.g.,14 processing and memory resources) while more accurately predicting the direct relation paths than prior techniques as evidenced by a higher mean reciprocal rank than prior techniques. Furthermore, in this manner, there is an improvement in the technical field involving knowledge representation and reasoning. Hence, there is an improvement in the technology or technical field of knowledge representation and reasoning, such as performing knowledge graph embedding more accurately and efficiently than prior techniques, using the additional elements, either alone or in combination with the recited judicial exception. See M.P.E.P. §2106.05(a). That is, using the claimed invention of claims 1, 10 and 19, knowledge graph embedding is performed more accurately and efficiently than prior techniques. See, e.g., paragraph [0021] of Applicant's specification. The technique of the claimed invention of claims 1, 10 and 19 utilizes fewer resources (e.g., processing and memory resources) while more accurately predicting the direct relation paths than prior techniques as evidenced by a higher mean reciprocal rank than prior techniques. See Id.
In regards to this argument, while the applicant makes the argument that the improvement lies in the claimed limitations, it does not mean that the abstract idea has been brought into the practical application. The improvement lies in the abstract ideas of the claim limitation. The improvement cannot be the abstract ideas alone. The improvement must, as well, be obvious within the claim itself and be more than instructions on how to perform the method using generic components. Further, it should be noted that, according to 2106.05(a)(II), ‘To show that the involvement of a computer assists in improving the technology, the claims must recite the details regarding how a computer aids the method, the extent to which the computer aids the method, or the significance of a computer to the performance of the method. Merely adding generic computer components to perform the method is not sufficient. Thus, the claim must include more than mere instructions to perform the method on a generic component or machinery to qualify as an improvement to an existing technology. See MPEP § 2106.05(f) for more information about mere instructions to apply an exception.’. The examiner maintains that doing a prediction with a prediction model and training said model is not enough to implement into the practical application as a prediction model is a ‘generic computer component’. The improvement is being claimed as a part of a limitation that has been designated as an abstract idea and cannot be a part of the exception as per MPEP 2106.05(a) “To show that the involvement of a computer assists in improving the technology, the claims must recite the details regarding how a computer aids the method, the extent to which the computer aids the method, or the significance of a computer to the performance of the method. Merely adding generic computer components to perform the method is not sufficient. Thus, the claim must include more than mere instructions to perform the method on a generic component or machinery to qualify as an improvement to an existing technology. See MPEP § 2106.05(f) for more information about mere instructions to apply an exception.”
With regards to the applicant’s remarks regarding the 103 rejection in the non-final action, the applicant argues that the prior art does not teach the claims 1, 10, and 19. The examiner acknowledges this argument and has adjusted the prior art of CREED to teach the newly added amendment. Namely, paragraph 83 of CREED has been added to the mapping to show the training of the prediction model for the explicit purpose of link/relationship prediction. Further, paragraph 124 has been added to show that CREED uses a feature vector set for the prediction.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SKIELER A KOWALIK whose telephone number is (571)272-1850. The examiner can normally be reached 8-5.
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, Mariela D Reyes can be reached at (571)270-1006. 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.
/SKIELER ALEXANDER KOWALIK/Examiner, Art Unit 2142 /Mariela Reyes/Supervisory Patent Examiner, Art Unit 2142
Prosecution Insights