DETAILED ACTION
This communication is in response to the Applicant Arguments/Remarks filed 3/17/2026. Claims 1-9, 12-27 are pending in the application.
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Arguments
Applicant's arguments filed 3/17/2026 have been fully considered. Regarding the arguments on pages 18-19 in relating to the amended limitations including “only” among tokens…, please see the new combination of references cited below. The teachings of Fonseca is no longer applied in the current Office action.
Senthivel et al. teaches at col. 3:18-24: provide a service that can be utilized in a simple and straightforward manner by clients to automatically identify entities-such as types of medications, treatments, medical conditions, etc. - and optionally, relationships involving these entities with other detected classes of objects such as attributes or traits from unstructured text; col. 8:1-6; col. 24:5-12: a robotic device can include sensors to capture input data. A user device can retrieve the model data from the training model data store and store the model data in the robotic device. The robotic device can then perform an action (e.g., move forward, raise an arm, generate a sound, etc.) based on the resulting output; col. 12:7-23: unmapped attributes are those attributes that are unable to be "mapped" to a particular entity with a sufficient amount (e.g., threshold) of confidence, though the attribute itself was found with some threshold amount of confidence; col. 33:15-18: upload new or modified data from a local cache so that the primary store of data is maintained. Thus, new attributes are maintained.
col. 8: 16-25: the UTAS 112-via use of these ML models-may detect information in multiples classes (or "object types"), such as entities, categories, types, attributes, traits, etc. An entity generally refers to a textual reference to the name of relevant objects, such as people, treatments, medications, or medical conditions-for example, "Ibuprofen" may be an entity. A category generally refers to a generalized grouping to which a detected entity belongs, for example, "Ibuprofen" may be part of a MEDICATION category. An attribute generally refers to information related to a detected entity, such as a dosage of a medication-for example, "200 mg" is an attribute of an "Ibuprofen" entity; fig. 2; col. 11:13-49: determine that a number of attributes are all related to the "Sodium Chloride" entity "Infuse" is a "route or mode" attribute, "0.9%" is a strength attribute etc.; col. 12:7-23: unmapped attributes are those attributes that are unable to be "mapped" to a particular entity with a sufficient amount (e.g., threshold) of confidence (thus, no relation), though the attribute itself was found with some threshold amount of confidence. The combination of references does teach the argued limitations.
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) 1-8, 10-16, 19-27 is/are rejected under 35 U.S.C. 103 as being unpatentable over Senthivel et al. (US 11487942) in view of Minton et al. (US 20090319515) and further in view of S Nanal et al. (US 20200180148).
As per claim 1 and 27, Senthivel et al. teaches
a method for identifying an attribute of an entity by processing an unstructured natural language text for task automation (col. 3:18-24: provide a service that can be utilized in a simple and straightforward manner by clients to automatically identify entities-such as types of medications, treatments, medical conditions, etc. - and optionally, relationships involving these entities with other detected classes of objects such as attributes or traits from unstructured text; col. 8:1-6; col. 24:5-12: a robotic device can include sensors to capture input data. A user device can retrieve the model data from the training model data store and store the model data in the robotic device. The robotic device can then perform an action (e.g., move forward, raise an arm, generate a sound, etc.) based on the resulting output),
the method being performed by a computing system, the method comprising: recognizing one or more entities in an input text; and selecting an attribute of a first entity included in the one or more entities among tokens included in the input text (figs. 2: detect entities from the unstructured text; fig. 7; col. 2:39-54: segments the unstructured data input and identifies tokens in these segments, and provides the segmented data and token information to be used with different machine learning (ML) models trained to detect different entity types within the segments. The output from ones of the models may be sent to one or more other models trained to identify relationships between detected classes of objects, such as attributes and entities; col. 7:15-19: the TSE then tokenizes the segments to identify one or more "tokens" within each segment. A token may be a word or a grouping of characters,
and the tokenization may be performed by applying another set of rules that indicate where the segment is to be split; col. 11:30-32: each entity may include an array of attributes extracted that relate to the entity; col. 16:53-56: identifying tokens within the plurality of segments, which may be performed based on applying another one or more
rules to the unstructured text (e.g., each of the segments) to identify locations of each token);
performing a task is performed using the first entity and the attribute of the first entity (col. 2:39-49: receives unstructured data (e.g., unstructured text) from a client associated with the user, segments the unstructured data input and identifies tokens in these segments, and provides the segmented data and token information
to be used with different machine learning (ML) models trained to detect different entity types within the segments. The output from ones of the models may be sent to one or
more other models trained to identify relationships between detected classes of objects, such as attributes and entities; col. 4:62-64: a user may directly utilize a compute instance hosted by the provider network to perform a variety of computing tasks; col. 24:5-18: a robotic device can include sensors to capture input data. A user device can retrieve the model data from the training model data store and store the model data in the robotic device),
wherein the selecting of the attribute of the first entity includes selecting the attribute of the first entity only among tokens that do not include the recognized one or more entities (col. 12:7-23: unmapped attributes are those attributes that are unable to be "mapped" to a particular entity with a sufficient amount (e.g., threshold)
of confidence, though the attribute itself was found with some threshold amount of confidence; col. 29:33-34: the training data and the evaluation data do not have common data; col. 33:15-18: upload new or modified data from a local cache so that the primary store of data is maintained).
wherein the selecting of the attribute of the first entity only among tokens that do not include the recognized one or more entities includes: selecting a plurality of candidate attributes only among the tokens that do not include the recognized one or more entities, using a part of speech of each token (col. 2:39-47: receives unstructured data (e.g., unstructured text) from a client associated with the user, segments the unstructured data input and identifies tokens in these segments, and provides the segmented data and token information to be used with different machine learning (ML) models trained to detect different entity types within the segments. The output from ones of the models may be sent to one or more other models trained to identify relationships between detected classes of objects, such as attributes and entities; col. 12:7-23: unmapped attributes are those attributes that are unable to be "mapped" to a particular entity with a sufficient amount (e.g., threshold) of confidence, though the attribute itself was found with some threshold amount of confidence; col. 33:15-18: upload new or modified data from a local cache so that the primary store of data is maintained; col. 30:31-36: personal digital assistant (PDA), hybrid PDA/mobile phone, mobile phone, electronic book reader, set-top box, voice command device, camera, digital media player, and the like);
determining a relation class between each of the plurality of candidate attributes and the first entity, as only any one of the following three classes of relations or less: is-a, part-of, and no-relation; selecting the attribute of the first entity using the determined relation class (col. 8: 16-25: the UTAS 112-via use of these ML models-may detect information in multiples classes (or "object types"), such as entities, categories, types, attributes, traits, etc. An entity generally refers to a textual reference to the name of relevant objects, such as people, treatments, medications, or medical conditions-for example, "Ibuprofen" may be an entity. A category generally refers to a generalized grouping to which a detected entity belongs, for example, "Ibuprofen" may be part of a MEDICATION category. An attribute generally refers to information related to a detected entity, such as a dosage of a medication-for example, "200 mg" is an attribute of an "Ibuprofen" entity; fig. 2; col. 11:13-49: determine that a number of attributes are all related to the "Sodium Chloride" entity "Infuse" is a "route or mode" attribute, "0.9%" is a strength attribute etc.; col. 12:7-23: unmapped attributes are those attributes that are unable to be "mapped" to a particular entity with a sufficient amount (e.g., threshold) of confidence (thus, no relation), though the attribute itself was found with some threshold amount of confidence).
Even if Senthivel does not explicitly teach wherein the selecting of the attribute of the first entity includes selecting the attribute of the first entity only among tokens that do not include the recognized one or more entities.
Minton et al. teaches
wherein the selecting of the attribute of the first entity includes selecting the attribute of the first entity only among tokens that do not include the recognized one or more entities (para. 90: If the received record is not a strong match for any entity attributes, a new entity is created and the record is entered into the new entity. In this step, it has been determined that the received record is not a strong match with any entity, and that a new entity will be created for the received record: para. 72: entity updates can involve creating a new entity, merging two or more entities into fewer entities, and dividing an entity into two or more entities. This may be particularly appropriate when some attributes in the record are close to the entity and others are not, i.e., when there is a high match probability for some attributes and a low match probability for other attributes).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Senthivel and the selecting the attribute of the first entity among tokens of Minton et al. in order to effectively extract attributes of remaining tokens in the text that were not recognized as entities.
Senthivel and Minton et al. do not explicitly teach using a robotic process automation (RPA) technology; performing a RPA (robotic process automation) task using the first entity and the attribute of the first entity.
S Nanal et al. teaches said limitation at para. 14: an analytical robotic process automation (RPA) system receives data regarding the issues arising during the processing of an entity through an application stack and implements solutions to resolve the issues. The application stack includes a plurality of applications wherein each application in the application stack can be configured to execute one or more functions or tasks that affect the attributes of the entity.
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Senthivel, Minton et al. and the performing a robotic process automation (RPA) task using the first entity and the attribute of the first entity of S Nanal et al. in order to allow quick responses for various search criteria and/or provides a quick summary for all entities old/new issues/tasks resolved.
As per claim 2, Senthivel et al. teaches
wherein the recognizing of the one or more entities includes recognizing only any one type of entity of a plurality of predetermined types in the input text (col. 2: 39-45: the service receives unstructured data (e.g., unstructured text) from a client associated with the user, segments the unstructured data input and identifies tokens in these segments, and provides the segmented data and token information to be used with different machine learning (ML) models trained to detect different entity types within the segments; col. 11:55-56: the segment of text is correctly recognized as an attribute, a Text string, an array of Traits, a Type string, etc.)
As per claim 3, Senthivel et al. teaches
wherein the recognizing of only any one type of entity of the plurality of
predetermined types includes recognizing a quantity type of entity or a code type
of entity (col. 7: an ML model trained to identify a particular type of entity; col. 8:61-67:
the medical condition service may return "aching pain" as a DX_NAME type of entity
that is of a SYMPTOM trait type, as well as another entity of "chronic" that is of the ACUITY; col. 16:27-36: receiving a request to identify entities in unstructured text. The
request may be received at a web service endpoint of a provider network and may
include an identifier of whether a particular type of entity (e.g., personal health
information entities) or multiple entities are to be detected within the unstructured
text).
As per claim 4, Senthivel et al. teaches
wherein the recognizing of only any one type of entity of the plurality of predetermined types includes recognizing a quantity type of entity or a code type of entity (col. 10:26-36: the attributes may include one or more of a DOSAGE attribute representing an amount of medication ordered, a DURATION attribute representing how long the medication should be administered; fig. 4: 50 mgs, 0.2 mgs etc.), and
the method further comprises: retrieving an input field corresponding to the attribute of the first entity; inputting the first entity as a value of the retrieved input field (col. 7, 9: last paragraph: an unstructured text input may be "Patient is John Smith, a 48 year old teacher and resident of Seattle, Wash." and the PHI service 118C may return that "John Smith" is an entity of type NAME, "48" is an entity of type AGE, "teacher" is an entity of type PROFESSION, "Seattle, Wash." is an ADDRESS entity; col. 10:18-21: the medication service, in response to a request, may return information that may include some or all of two entity types, seven attributes, and one trait; col. 13:9-17).
As per claim 5, Senthivel et al. teaches
wherein the input text is a natural language text included in a medical record, and
the input field is included in one of a plurality of input forms belonging to an electronic medical record (EMR) (col. 3:53-62: detect useful medical-related information in unstructured text such as clinical text. As much as 75% of all health
record data is found in unstructured text, e.g., in physician's notes, discharge summaries, test results, case notes, and so on, the UTAS can utilize uses Natural Language Processing (NLP) models to sort through this enormous quantity of data and retrieve valuable information; col. 6:9-15: client may be part of a medical records application, e.g., a medical billing system payor can use the UTAS to expand its analytics to include the use of unstructured documents such as clinical notes, where more information about a diagnosis as it relates to billing codes can be determined; col. 10:58-67).
As per claim 6, Senthivel et al. teaches
generating training data, the training data comprising entity-attribute pairs, each entity-attribute pair including a corresponding entity of the one or more entities and an attribute of the corresponding entity (col. 8:1-33: an attribute generally refers to information related to a detected entity, such as a dosage of a medication-for example, "200 mg" is an attribute of an "Ibuprofen" entity; col. 10:40-67: utilize one or more relationship models 124A-124N to detect relationships between these entities (or other types of information) at circle (6). These relationship models may be, for example,
neural networks such as Convolutional Neural Networks (CNNs) trained with labeled training data indicating relationships between entities and attributes, etc., by detecting that an attribute of "80 mg" and an attribute of "daily" is associated with an entity of "Aspirin" in the unstructured text "The patient has been daily taking 80 mg of Aspirin").
As per claim 7, Senthivel et al. teaches
wherein the entity-attribute pair further includes a sentence including the corresponding entity, a type of the corresponding entity, and a relation class between the corresponding entity and the attribute (col. 2:45-48: the output from ones of the models may be sent to one or more other models trained to identify relationships between detected classes of objects, such as attributes and entities; col. 10:18-29: the attributes may include one or more of a DOSAGE attribute representing an amount of medication ordered, a DURATION attribute representing how long the
medication should be administered).
As per claim 8, Senthivel et al. teaches
wherein the selecting of the attribute of the first entity only among the tokens that do not include the recognized one or more entities includes: segmenting the input text into a plurality of unit texts and selecting the attribute of the first entity among tokens that is included in a unit text including the first entity and do not include the recognized one or more entities (fig. 1: tokenization and segmentation engine; fig. 7: identify a plurality of segments within the unstructured text; col. 2:40-45: segments the unstructured data input and identifies tokens in these segments, and provides the segmented data and token information to be used with different machine learning (ML) models trained to detect different entity types within the segments; col. 7:15-33: segment "Infuse Sodium Chloride 0.9% solution", a token of "Infuse" may be identified via a beginning offset of "0" and an ending offset of "6," or via a beginning offset of"0" and a length of "6."; col. 12:7-23: unmapped attributes are those attributes that are unable to be "mapped" to a particular entity with a sufficient amount (e.g., threshold) of confidence, though the attribute itself was found with some threshold amount of confidence; col. 29:33-34: the training data and the evaluation data do not have common data; col. 33:15-18: upload new or modified data from a local cache so that the primary store of data is maintained)
Even if Senthivel does not explicitly teach selecting of the attribute of the first entity only among the tokens that do not include the recognized one or more entities.
Minton et al. teaches
wherein the selecting of the attribute of the first entity includes selecting the attribute of the first entity only among tokens that do not include the recognized one or more entities (para. 90: If the received record is not a strong match for any entity attributes, a new entity is created and the record is entered into the new entity. In this step, it has been determined that the received record is not a strong match with any entity, and that a new entity will be created for the received record: para. 72: entity updates can involve creating a new entity, merging two or more entities into fewer entities, and dividing an entity into two or more entities. This may be particularly appropriate when some attributes in the record are close to the entity and others are not, i.e., when there is a high match probability for some attributes and a low match probability for other attributes).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Senthivel and the selecting the attribute of the first entity among tokens of Minton et al. in order to effectively extract attributes of remaining tokens in the text that were not recognized as entities.
As per claim 12, Senthivel et al. teaches
wherein the determining of the relation class includes: determining a plurality of relation classes corresponding to a type of the first entity; and determining the relation class between each of the plurality of candidate attributes and the first entity as any one of the determined relation classes (col. 2:43-49: detect different entity types within the segments. The output from ones of the models may be sent to one or more other models trained to identify relationships between detected classes of objects, such as attributes and entities. The outputs may then be consolidated and returned to the client in a unified response; col. 8:16-25: the UTAS 112-via use of these ML models-may detect information in multiples classes (or "object types"), such as entities, categories, types, attributes, traits, etc. An entity generally refers to a textual reference to the name of relevant objects, such as people, treatments, medications, or medical conditions-for example, "Ibuprofen" may be an entity. A category generally refers to a generalized grouping to which a detected entity belongs, for example, "Ibuprofen" may be part of a MEDICATION category. An attribute generally refers to information related to a detected entity, such as a dosage of a medication-for example, "200 mg" is an attribute of an "Ibuprofen" entity; fig. 2; col. 11:13-49: as shown in the visual representation shown in FIG. 2, this may result in the orchestrator being able to determine that a number of attributes are all related to the "Sodium Chloride" entity "Infuse" is a "route or mode" attribute, "0.9%" is a strength attribute etc.).
As per claim 13, Senthivel et al. teaches
wherein the determining of the plurality of relation classes corresponding to the type of the first entity includes: determining at least one of two relation classes: is-a and part-of as some of the plurality of relation classes corresponding to the type of the first entity (col. 9: last paragraph: an unstructured text input may be "Patient is John Smith, a 48 year old teacher and resident of Seattle, Wash." and the PHI service 118C may return that "John Smith" is an entity of type NAME, "48" is an entity of type AGE, "teacher" is an entity of type PROFESSION, "Seattle, Wash." is an ADDRESS entity; col. 10:18-21: the medication service, in response to a request, may return information that may include some or all of two entity types, seven attributes, and one trait; col. 13:9-17; col. 8:16-25: the UTAS 112-via use of these ML models-may detect information in multiples classes (or "object types"), such as entities, categories, types, attributes, traits, etc. An entity generally refers to a textual reference to the name of relevant objects, such as people, treatments, medications, or medical conditions-for example, "Ibuprofen" may be an entity. A category generally refers to a generalized grouping to which a detected entity belongs, for example, "Ibuprofen" may be part of a MEDICATION category. An attribute generally refers to information related to a detected entity, such as a dosage of a medication-for example, "200 mg" is an attribute of an "Ibuprofen" entity); and
determining a class: no-relation as the other of the plurality of relation classes
corresponding to the type of the first entity (col. 2:43-49: detect different entity types within the segments. The output from ones of the models may be sent to one or more other models trained to identify relationships between detected classes of objects, such as attributes and entities. The outputs may then be consolidated and returned to the client in a unified response; col. 9:42-67: detect a variety of different types of entities, including but not limited to an AGE type that represents components of age, spans of age or other age mentioned in the unstructured text, a NAME type that represents names mentioned in the text, typically belonging to a patient, family, or provider, a PHONE_OR_FAX type that represents phone numbers or FAX numbers (and may eliminate certain named phone numbers, such as 1-800-QUIT-NOW or 911), an EMAIL type that represents email addresses, an ID type that represents a social security
number, medical record number, facility identification number).
As per claim 14, Senthivel et al. teaches
wherein the determining the relation class between each of the plurality of candidate attributes and the first entity includes selecting a candidate attribute having a relation class corresponding to a type of the entity as the attribute of the entity (col. 2:32-49: relationship detection from unstructured text. The output from ones of the models may be sent to one or more other models trained to identify relationships between detected classes of objects, such as attributes and entities. The outputs may then be consolidated and returned to the client in a unified response; col. 8:15-67: an entity generally refers to a textual reference to the name of relevant objects, such as people, treatments, medications, or medical conditions-for example, "Ibuprofen" may be an entity. A category generally refers to a generalized grouping to which a detected entity belongs, for example, "Ibuprofen" may be part of a MEDICATION category, …, the medical condition service may return "aching pain" as a DX_NAME type of entity that is of a SYMPTOM trait type, as well as another entity of "chronic" that is of the ACUITY).
As per claim 15, Senthivel et al. teaches
wherein the determining of the relation class includes: determining the relation class between each of the plurality of candidate attributes and the first entity using a first relation extraction model when a type of the first entity is a first type (col. 7:46-50: each ML model may be implemented as part of a service (or “micro-service”) that receives inference requests, optionally pre-processes the input data, provides the provided input (or pre-processed input) to an ML model trained to identify a particular type of entity; col. 11:1-23); and
determining the relation class between each of the plurality of candidate attributes and the first entity using a second relation extraction model different: from the first relation extraction model when the type of the first entity is a second type different from the first type, the first relation extraction model and the second relation extraction model are models trained based on machine learning (col. 2:50-54: the service may implement various ML models trained to detect medical-related entities from unstructured text (such as doctors' notes). Clients may call the service to request the identification of any entities that can be found, or just specific entities; col. 11:12-21: identify which attributes correspond to which entities. As shown in the visual representation 210 shown in FIG. 2, this may result in the orchestrator being able to determine that a number of attributes are all related to the "Sodium Chloride" entity "Infuse" is a "route or mode" attribute, "0.9%" is a strength attribute, "solution" is a form attribute, " 1000 mL" is a dosage attribute, "Intravenously" is a route or mode attribute, "daily" is a frequency attribute, and "3 days" is a duration attribute),
receiving input data including a sentence, an entity, and an attribute, and outputting data related to what a relation between the entity and the attribute belongs to any one of a plurality of relation classes (col. 9:61-67: an unstructured text input may be “Patient is John Smith, a 48 year old teacher and resident of Seattle, Wash.” and the PHI service 118C may return that “John Smith” is an entity of type NAME, “48” is an entity of type AGE, “teacher” is an entity of type PROFESSION, “Seattle, Wash.” is an ADDRESS entity),
the first relation extraction model outputs data related to what the relation between the entity and the attribute belongs to any one of a plurality of first relation classes, the second relation extraction model outputs data related to what the relation between the entity and the attribute belongs to any one of a plurality of second relation classes (col. 13:12-17: a collection of the medical entities extracted from the input text and their associated information. For each entity, the response provides the entity text, the entity category, where the entity text begins and ends, and the level of confidence in the detection and analysis. Attributes and traits of the entity are also returned; col. 2:32-47: provides the segmented data and token information to be used with different machine learning (ML) models trained to detect different entity types within the segments. The output from ones of the models may be sent to one or more other models trained to identify relationships between detected classes of objects, such as attributes and entities), and
at least one of the plurality of first relation classes includes one or more first noncommon relation classes which is not included in the plurality of second relation classes and at least one of the plurality of second relation classes includes one or more second non-common relation classes which is not included in the plurality of first relation classes (col. 3:20-24: automatically identify entities—such as types of medications, treatments, medical conditions, etc. —and optionally, relationships involving these entities with other detected classes of objects such as attributes or traits—from unstructured text; col. 29:33-34: the training data and the evaluation data do not have common data).
As per claim 16, Senthivel et al. teaches
wherein the selecting of the plurality of candidate attributes includes excluding some of the plurality of candidate attributes from the plurality of candidate attributes using a relation between each of the plurality of candidate attributes and the entity; the selecting of the attribute of the first entity using the determined relation class (col. 10:54-56: identify relationships between the detected information - e.g., which attributes belong to (or, are associated with) which entities; col. 2:43-49: different machine learning (ML) models trained to detect different entity types within the segments. The output from ones of the models may be sent to one or more other models trained to identify relationships between detected classes of objects, such as attributes and entities. The outputs may then be consolidated and returned to the
client in a unified response; col. 8:17-33).
Senthivel does not explicitly teach excluding some of the plurality of candidate attributes from the plurality of candidate attributes using a relation between each of the plurality of candidate attributes and the entity; selecting the attribute of the first entity among the candidate attributes remaining after the excluding of some of the plurality of candidate attributes, using the token distance of each of the candidate attributes,
Minton et al. teaches said limitations at para. 78: The "blocking phase" as discussed with respect to the method of FIG. 5 is to very quickly identify the most promising candidates from a much larger set of possible candidates. Blocking may rely on simple yet efficient techniques for reducing the space of possible candidates, for example by using token-based distance metrics (Jaccard similarity coefficients, term frequency-inverse document frequency [TF-IDF], etc.; para. 35: An entity knowledgebase may be created and applied for just about any type of entity, including people, organizations, companies, terrorist groups, and so on. An entity knowledgebase could also be used to process data in a database or to reason about the relationships between entities; para. 55: the coarse-grain option, on the other hand, may not identify the specific attribute that matches, but may eliminate the need to unambiguously parse the data. If the data is treated as a sequence of tokens, i.e., a document, there may be no need to resolve ambiguous parses when storing the data. However, this may mean that the data must later be parsed at run-time or "on the fly," possibly producing sub-optimal performance)..
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Senthivel and the selecting the attribute of the first entity among tokens of Minton et al. in order to effectively extract attributes of remaining tokens in the text that were not recognized as entities.
As per claim 19, Senthivel et al. teaches
wherein the selecting of the attribute of the first entity only among tokens that do not include the recognized one or more entities further includes (col. 29:33-34: the training data and the evaluation data do not have common data; col. 10:54-56: identify relationships between the detected information - e.g., which attributes belong to (or, are associated with) which entities; col. 2:45-47: the output from ones of the models may be sent to one or more other models trained to identify relationships between detected classes of objects, such as attributes and entities; col. 30: line 35: voice command device; col. 12:7-23: unmapped attributes are those attributes that are unable to be "mapped" to a particular entity with a sufficient amount (e.g., threshold) of confidence, though the attribute itself was found with some threshold amount of confidence; col. 33:15-18: upload new or modified data from a local cache so that the primary store of data is maintained);
determining a token distance between each of the plurality of candidate attributes and the first entity; and selecting the attribute of the first entity by partially using the token distance of each of the candidate attributes (col. 23:15-21: the ML model evaluator can then compare the outputs of the ML model to the expected outputs and determine one or more quality metrics of the ML model being trained based on the comparison (e.g., the error rate can be a difference or distance between the ML model outputs and the expected outputs); col. 22:2-7: deploy and/or execute a version of a partially trained ML model (e.g., an ML model trained as of a certain stage in the training process). A version of a partially-trained ML model can be based on some or all of the model data files stored in the training model data store).
Even if Senthivel does not explicitly teach wherein the selecting of the attribute of the first entity includes selecting the attribute of the first entity only among tokens that do not include the recognized one or more entities,
Minton et al. teaches said limitations at para. 90: If the received record is not a strong match for any entity attributes, a new entity is created and the record is entered into the new entity. In this step, it has been determined that the received record is not a strong match with any entity, and that a new entity will be created for the received record: para. 72: entity updates can involve creating a new entity, merging two or more entities into fewer entities, and dividing an entity into two or more entities. This may be particularly appropriate when some attributes in the record are close to the entity and others are not, i.e., when there is a high match probability for some attributes and a low match probability for other attributes).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Senthivel and the selecting the attribute of the first entity among tokens of Minton et al. in order to effectively extract attributes of remaining tokens in the text that were not recognized as entities.
As per claim 20, Senthivel et al. teaches
wherein the selecting of the attribute of the first entity further includes determining a relation class between each of the plurality of candidate attributes and the first entity (col. 10:54-56: identify relationships between the detected information - e.g., which attributes belong to (or, are associated with) which entities; col. 2:45-47: the output from ones of the models may be sent to one or more other models trained to identify relationships between detected classes of objects, such as attributes and entities),
and the selecting of the attribute of the first entity by partially using the token distance of each of the candidate attributes includes selecting the attribute of the first entity using the token distance of each of the candidate attributes and the determined relation class (col. 23:15-21: a level of confidence that the accuracy of the ML model being trained is known, etc. The ML model evaluator can obtain the model data for an ML model being trained and evaluation data from the training data store. The evaluation data is separate from the data used to train an ML model and includes both input data and expected outputs (e.g., known results), and thus the ML model evaluator can define an ML model using the model data and execute the ML model by providing the input data as inputs to the ML model. The ML model evaluator can then compare the outputs of the ML model to the expected outputs and determine one or more quality metrics of the ML model being trained based on the comparison (e.g., the error rate can be a difference or distance between the ML model outputs and the expected outputs)).
As per claim 21, Senthivel et al. teaches
wherein the selecting of the attribute of the first entity only among tokens that do not include the recognized one or more entities further includes (col. 29:33-34: the training data and the evaluation data do not have common data; col. 10:54-56: identify relationships between the detected information - e.g., which attributes belong to (or, are associated with) which entities; col. 2:45-47: the output from ones of the models may be sent to one or more other models trained to identify relationships between detected classes of objects, such as attributes and entities; col. 30: line 35: voice command device; col. 12:7-23: unmapped attributes are those attributes that are unable to be "mapped" to a particular entity with a sufficient amount (e.g., threshold) of confidence, though the attribute itself was found with some threshold amount of confidence; col. 33:15-18: upload new or modified data from a local cache so that the primary store of data is maintained);
retrieving a record for each of the plurality of candidate attributes from a pre-stored statistical table (col. 23:2-8: the model metrics can include quality metrics, such as an error rate of the ML model being trained, a statistical distribution of the ML model being trained, a latency of the ML model being trained, a confidence level of the ML model being trained (e.g., a level of confidence that the accuracy of the ML model being trained is known, etc.); and
selecting the attribute of the first entity using the retrieved record of each of the plurality of candidate attributes, the pre-stored statistical table includes a record of each attribute, the record includes a type of an entity, the number of times of extraction of the entity, information on a distance between an attribute and the entity, and a confidence score, and the retrieved record is a record of a candidate attribute having a type of an entity coinciding with a type of the first entity (col. 5, line 58 to col. 6, line 8: a client may operate as part of a clinical research application allowing life sciences or research organizations to optimize the matching process for fitting patients into clinical trials using information from unstructured clinical texts, such as case notes and test results; col. 11:7-13; col. 23:15-21: a level of confidence that the accuracy of the ML model being trained is known, etc. The ML model evaluator can obtain the model data for an ML model being trained and evaluation data from the training data store. The evaluation data is separate from the data used to train an ML model and includes both input data and expected outputs (e.g., known results), and thus the ML model evaluator can define an ML model using the model data and execute the ML model by providing the input data as inputs to the ML model. The ML model evaluator can then compare the outputs of the ML model to the expected outputs and determine one or more quality metrics of the ML model being trained based on the comparison (e.g., the error rate can be a difference or distance between the ML model outputs and the expected outputs).
Even if Senthivel does not explicitly teach selecting a plurality of candidate attributes among the tokens that do not include the recognized one or more entities using a part of speech of each token.
Minton et al. teaches said limitations at para. 90: f the received record is not a strong match for any entity attributes, a new entity is created and the record is entered into the new entity. In this step, it has been determined that the received record is not a strong match with any entity, and that a new entity will be created for the received record: para. 72: entity updates can involve creating a new entity, merging two or more entities into fewer entities, and dividing an entity into two or more entities. This may be particularly appropriate when some attributes in the record are close to the entity and others are not, i.e., when there is a high match probability for some attributes and a low match probability for other attributes; para. 78: The "blocking phase" as discussed with respect to the method of FIG. 5 is to very quickly identify the most promising candidates from a much larger set of possible candidates. Blocking may rely on simple yet efficient techniques for reducing the space of possible candidates, for example by using token-based distance metrics (Jaccard similarity coefficients, term frequency-inverse document frequency [TF-IDF], etc.)
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Senthivel and the selecting the attribute of the first entity among tokens of Minton et al. in order to effectively extract attributes of remaining tokens in the text that were not recognized as entities.
As per claim 22, Senthivel et al. teaches
wherein the selecting of the attribute of the first entity using the retrieved record of each of the plurality of candidate attributes includes: calculating a difference between a record of the retrieved record of each of the plurality of candidate attributes and a distance between the first entity and a first candidate attribute on the input text (col. 23:15-21: the ML model evaluator can then compare the outputs of the ML model to the expected outputs and determine one or more quality metrics of the ML model being trained based on the comparison (e.g., the error rate can be a difference or distance between the ML model outputs and the expected outputs));
adjusting the confidence score of the retrieved record for each of the plurality of
candidate attributes using the calculated difference of each of the plurality of candidate attributes; and selecting the attribute of the first entity using the adjusted confidence score of each of the plurality of candidate attributes (col. 11:7-13; col. 23:21-42: the ML model evaluator periodically generates model metrics during the training process and stores the model metrics in the training metrics data store. While the ML model is being trained, a user, via the user device, can access and retrieve the model metrics from the training metrics data store. The user can then use the model metrics to determine whether to adjust the training process and/or to stop the training process. For example, the model metrics can indicate that the ML model is performing poorly (e.g., has an error rate above a threshold value, has a statistical distribution that is not an expected or desired distribution (e.g., not a binomial distribution, a Poisson distribution, a geometric distribution, a normal distribution, Gaussian distribution, etc.), has an execution latency above a threshold value, has a confidence level below a threshold value)) and/or is performing progressively worse (e.g., the quality metric continues to worsen over time). In response, in some embodiments, the user, via the user device, can transmit a request to the model training system to modify the ML model being trained).
As per claim 23, Senthivel et al. teaches
wherein the record further includes information on a relation class between the attribute and the entity, and the retrieved record is a record of the candidate attribute having a relation class value coinciding with a relation class between the first entity and the candidate attribute (col. 5, line 58 to col. 6, line 8: a client may operate as part of a clinical research application allowing life sciences or research organizations to optimize the matching process for fitting patients into clinical trials using information from unstructured clinical texts, such as case notes and test results).
As per claim 24, Senthivel et al. teaches
wherein the selecting of the attribute of the first entity using the retrieved record of each of the plurality of candidate attributes includes: calculating a confidence score of a first candidate attribute using a token distance between the first candidate attribute and the first entity and a relation class between the first candidate attribute and the first entity when a record corresponding to the first candidate attribute of the plurality of candidate attributes is not retrieved from the pre-stored statistical table (col. 23:10-21: the evaluation data is separate from the data used to train an ML model and includes both input data and expected outputs (e.g., known results), and thus the ML model evaluator can define an ML model using the model data and execute the ML model by providing the input data as inputs to the ML model. The ML model evaluator can then compare the outputs of the ML model to the expected outputs and determine one or more quality metrics of the ML model being trained based on the comparison (e.g., the error rate can be a difference or distance between the ML model outputs and the expected outputs); and
selecting the attribute of the first entity by comparing the calculated confidence score with a confidence score of the retrieved record (col. 3, last paragraph: provide confidence scores that indicate the level of confidence in the accuracy of the detected entities, which can be used to enable client systems to apply more (or less) scrutiny to its results based on the particular use case; col. 8:last paragraph: the medical condition service may also provide associated confidence scores generated by the model when detecting each entity).
As per claim 25, Senthivel et al. teaches
wherein the selecting of the attribute of the first entity only among the tokens that do not include the recognized one or more entities further includes (col. 29:33-34: the training data and the evaluation data do not have common data; col. 10:54-56: identify relationships between the detected information - e.g., which attributes belong to (or, are associated with) which entities; col. 2:45-47: the output from ones of the models may be sent to one or more other models trained to identify relationships between detected classes of objects, such as attributes and entities; col. 30: line 35: voice command device; col. 12:7-23: unmapped attributes are those attributes that are unable to be "mapped" to a particular entity with a sufficient amount (e.g., threshold) of confidence, though the attribute itself was found with some threshold amount of confidence; col. 33:15-18: upload new or modified data from a local cache so that the primary store of data is maintained);
constructing input data for each of the plurality of candidate attributes, the input data including a target text, the first entity, a candidate attribute, and a relation class between the first entity and the candidate attribute (col. 5:47-57: manage their unstructured notes effectively and rapidly assess medical information about their patients that doesn't easily fit into the forms traditionally used. Analyzing case notes, for instance, may help providers identify candidates for early screening for certain medical conditions before the condition becomes more difficult to treat. It may also allow patients to report their health concerns in a narrative that can provide more information in a simple format, and then make those narratives easily available to providers in a more structured form, allowing more accurate diagnosis of medical conditions); and
inputting the input data for each of the plurality of candidate attributes into a pre-trained deep learning-based attribute identification model and selecting the attribute of the first entity among the plurality of candidate attributes using data output from the pre-trained deep learning based attribute identification model (col. 13:47-58: a first result user interface is illustrated showing different detected entities, attributes, etc. In this example, various aspects (e.g., entities, attributes, traits, etc.) are shown with underlines—which may be colorized to reflect which category (e.g., MEDICATION, MEDICAL_CONDITION, PROTECTED_HEALTH_INFORMATION, TEST_TREATMENT_PROCEDURE, ANATOMY) the information is. Relationships that are detected may also be shown with arrows linking an attribute to the entity, and in this example the arrows are labeled with the particular attribute identifier of the attribute (e.g., “0.2 mgs” is a strength attribute of the entity “Clonidine”)).
Even if Senthivel does not explicitly teach wherein the selecting of the attribute of the first entity includes selecting the attribute of the first entity only among tokens that do not include the recognized one or more entities,
Minton et al. teaches said limitations at para. 90: If the received record is not a strong match for any entity attributes, a new entity is created and the record is entered into the new entity. In this step, it has been determined that the received record is not a strong match with any entity, and that a new entity will be created for the received record: para. 72: entity updates can involve creating a new entity, merging two or more entities into fewer entities, and dividing an entity into two or more entities. This may be particularly appropriate when some attributes in the record are close to the entity and others are not, i.e., when there is a high match probability for some attributes and a low match probability for other attributes).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Senthivel and the selecting the attribute of the first entity among tokens of Minton et al. in order to effectively extract attributes of remaining tokens in the text that were not recognized as entities.
As per claim 26, Senthivel et al. teaches
wherein the pre-trained deep learning-based attribute identification model is generated through additional training using training data comprising the target text, an entity, an attribute, a relation class between the entity and the attribute, based on a deep learning-based base model performing a relation extraction (RE) task between the entities (col. 7, last para.: each ML model may be implemented as part of a service (or “micro-service”) that receives inference requests, optionally pre-processes the input data, provides the provided input (or pre-processed input) to an ML model trained to identify a particular type of entity, optionally post-processes the output inference result, and returns the (optionally post-processed) inference result to the client—here, the orchestrator; col. 28:47-50: the operating environment supports many different types of ML models, such as multi-arm bandit models, reinforcement learning models, ensemble ML models, deep learning models, or the like).
Claim(s) 9 is/are rejected under 35 U.S.C. 103 as being unpatentable over Senthivel et al. (US 11487942) in view of Minton et al. (US 20090319515) and further in view of S Nanal et al. (US 20200180148) and Carus (US 5890103).
As per claim 9, Senthivel et al. does not explicitly teach claim 9.
Carus teaches
wherein the selecting of the attribute of the first entity among the tokens that is included in the unit text including the first entity and do not include the recognized one or more entities includes: skipping the selecting of the attribute for a unit text in which any one type of named entity of a plurality of predetermined types is not recognized among the plurality of unit texts (col. 2:33-37: this new tokenization paradigm allows the invention to associate the attributes of lexical matter found in a token and the attributes of non-lexical matter following the token with the token; col. 13:49-64: tokenizer 43 extracts tokens (i.e., white-space delimited strings with leading and trailing punctuation removed) from a stream of natural language text; col. 17:26-43: the disambiguator module forms a window of sequential tokens containing will run after skipping those words having ignore tags in the following phrases: will run; will frequently run; will very frequently run; will not run; and will never run. The second embodiment thus ensures, by skipping or ignoring a class of irrelevant tokens, an accurate and rapid contextual analysis of the ambiguous token without having to expand the number of tokens in the window of sequential tokens. Moreover, a window of four sequential tokens ranging from the two tokens immediately preceding the ambiguous token and the token immediately following the ambiguous token can be expanded to include additional tokens by: (1) skipping those tokens contained within the original window of four sequential tokens that have ignore tags, and (2) replacing the skipped tokens with additional sequential tokens surrounding the ambiguous token).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Senthivel, Minton, S Nanal et al. and skipping the selecting of the attribute Carus in order to effectively prevents wrong data and avoids system or machine learning models errors.
Claim(s) 17-18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Senthivel et al. (US 11487942) in view of Minton et al. (US 20090319515) and further in view of S Nanal et al. (US 20200180148) and Vashist et al. (US 20220253729).
As per claim 17, Senthivel, Minton, S Nanal et al. do not explicitly teach claim 17,
Vashist et al. teaches
wherein the selecting of the attribute of the first entity among the candidate attributes remaining after the excluding of some of the plurality of candidate attributes, using the token distance of each of the candidate attributes includes: determining a context distance between the candidate attribute and the first entity (para. 37-40: extract values for the attributes associated with each of the entities. In some cases, the entities/attributes may be stored in a data structure as slot-value pairs, entity extraction subsystem may implement an information model, such as information model, to perform the entity/attribute/value extraction. Some embodiments include information model used by NLP entity extraction models to perform feature extraction; para. 110 - A6: determining, based on each distance, that the structured data records are classified as being similar to a respective document comprising the respective unstructured data; fig. 4: entity extraction subsystem, tokenization module, sentence splitter module);
selecting the attribute of the first entity among the candidate attributes remaining after the excluding of some of the plurality of candidate attributes, using the token distance and the context distance of each of the candidate attributes (para. 3: matching the structured data record and the document based on a common domain of interest and extracting features from the unstructured data based on a natural language processing (NLP) entity extraction model that tokenizes the unstructured data and uses domain-specific entity identification of the tokenized unstructured data; para. 41, 110 – A5: generating a first feature vector representing text included in a given sentence; mapping the first feature vector to a coordinate location in a multidimensional feature space; and determining a group of feature vectors having a distance from the coordinate location that is less than a distance threshold, wherein the sentences that are identified comprise sentences whose feature vectors map to coordinate locations in the multidimensional feature space that is less than the distance threshold; para. 55).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Senthivel, Minton, S Nanal et al. and the using the token distance of each of the candidate attributes of Vashist et al. in order to effectively extract attributes of remaining tokens in the text that are recognized as relating to entities.
As per claim 18, Senthivel et al. does not explicitly teach said claim.
Minton et al. teaches said limitations at para. 78: The "blocking phase" as discussed with respect to the method of FIG. 5 is to very quickly identify the most promising candidates from a much larger set of possible candidates. Blocking may rely on simple yet efficient techniques for reducing the space of possible candidates, for example by using token-based distance metrics (Jaccard similarity coefficients, term frequency-inverse document frequency [TF-IDF], etc.; para. 35: An entity knowledgebase may be created and applied for just about any type of entity, including people, organizations, companies, terrorist groups, and so on. An entity knowledgebase could also be used to process data in a database or to reason about the relationships between entities; para. 55: the coarse-grain option, on the other hand, may not identify the specific attribute that matches, but may eliminate the need to unambiguously parse the data. If the data is treated as a sequence of tokens, i.e., a document, there may be no need to resolve ambiguous parses when storing the data. However, this may mean that the data must later be parsed at run-time or "on the fly," possibly producing sub-optimal performance).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teachings of Senthivel and the selecting the attribute of the first entity among tokens of Minton et al. in order to effectively extract attributes of remaining tokens in the text that were not recognized as entities.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. Goldenberg et al. (US 20090089630) teaches at para. 128: Perform matched candidate pairs reduction [0129] Generate matched set, matched statistics, and initial weights [0130] Skip last step because of too few attributes [0131] Iterate over previous step and check for convergence of weights [0132] Execute all remaining steps through end of process.
Kummamuru et al. (US 20190370397) teaches said limitation at para. 27: the rules 164 can be more complex or granular so that the rules could enable selection of entities or phrases. For example, there can be words or phrases which refer to supplier names as well as names of people.
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to LINH BLACK whose telephone number is (571)272-4106. The examiner can normally be reached 9AM-5PM EST M-F.
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, Tony Mahmoudi can be reached at 571-272-4078. 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.
/LINH BLACK/Examiner, Art Unit 2163 7/30/2026
/TONY MAHMOUDI/Supervisory Patent Examiner, Art Unit 2163
Prosecution Insights