Semi-supervised system for domain-specific emotion learning

Abstract

Automated computer systems and methods for determining information sentiments in digital information or content are disclosed. One aspect includes deriving, by a processor, the digital information from a source; generating, by the processor, a domain-specific machine learning emotion score through one of at least two machine learning models based on the digital information; autonomously drawing, by the processor, a non-domain-specific knowledge graph of associations between elements in a set of digital context information; receiving, by the processor, emotional maps, each emotional map defining an emotion; generating, by the processor, a map mood score based on the non-domain-specific knowledge map and the mood map; generating, by the processor, a final emotion score based on the map emotion score and the domain-specific machine learning emotion score; and determining the emotion of the information in the digital information via the final emotion score.

Description

[0001] Cross-reference to related applications

[0002] This application claims the benefit of and priority under 35 U.S.C. §119(e) to U.S. Provisional Application No. 63 / 377,994, filed on September 20, 2023, entitled “SEMI-SUPERVISED SYSTEM FOR DOMAIN SPECIFIC SENTIMENT LEARNING,” the contents of which are hereby incorporated by reference in their entirety. Technical Field

[0003] The present technology relates to systems and methods for training and enhancing machine learning networks to determine the sentiment or tone of digital information, news items, and other information sources related to a specific domain. Specifically, but not by way of limitation, the present technology relates to systems and methods for semi-supervised domain-specific sentiment learning. Background Art

[0004] Training machine learning networks using semi-supervised techniques produces outputs of varying usability and usefulness, particularly in the information and news domain and in the area of ​​sentiment classification. There is a need to improve the results of semi-supervised training models and enhance their outputs to generate more accurate and consistently usable results. Summary of the invention

[0005] In various aspects, the present disclosure provides a method for determining information sentiment in digital information, the method comprising: exporting the digital information from a source by a processor; generating a domain-specific machine learning sentiment score by the processor based on the digital information through one of at least two machine learning models; autonomously drawing a non-domain-specific knowledge graph of associations between elements in a set of digital background information by the processor; receiving a sentiment graph by the processor, each sentiment graph defining a sentiment; generating a graph sentiment score by the processor based on the non-domain-specific knowledge graph and the sentiment graph; generating a final sentiment score by the processor based on the graph sentiment score and the domain-specific machine learning sentiment score; and determining the information sentiment in the digital information by the processor via the final sentiment score.

[0006] In various aspects, the method may also include automatically updating entity attributes in at least one of a database or a server based on the final sentiment score.

[0007] In various aspects, the method includes: training a first machine learning model having a base layer and a second layer for the digital information; incorporating the base layer trained for the digital information into a second machine learning model; and training the second machine learning model including the base layer and the final layer to generate the domain-specific machine learning sentiment score. In some aspects, the training of the first machine learning model includes training the first machine learning model to classify the subject matter of the digital information.

[0008] In various aspects, the generation of the graph emotion score includes: determining the graph similarity between each emotion graph in the emotion graph and the non-domain-specific knowledge graph; applying the emotions defined by each emotion graph in the emotion graph to its determined graph similarity to generate a graph-specific similarity tone score; and combining the graph-specific similarity tone scores of the emotion graphs.

[0009] In various aspects, the generation of the final sentiment score includes: applying a weighting to the graph sentiment score to generate a weighted graph sentiment score; applying another weighting to the domain-specific machine learning sentiment score to generate a weighted domain-specific machine learning sentiment score; and combining the weighted graph sentiment score and the weighted domain-specific machine learning sentiment score.

[0010] In the method, the element may also include at least one of an entity, a name, a location, a time, or an event. The method may also include at least a portion of the digital information being labeled. In some aspects of the method, the emotion defined by each emotion map is related to a background context provided in a digital manner.

[0011] In various aspects, the present disclosure provides an automated system for defining information sentiment in digital information, the system comprising: at least one of a database or a server containing entity attributes; a processor; and a computer-readable medium storing instructions, the instructions being executable by the processor to perform the following operations: the processor derives the digital information from a source; the processor generates a domain-specific machine learning sentiment score based on the digital information through one of at least two machine learning models; the processor autonomously draws a non-domain-specific knowledge graph of associations between elements in a set of digital background information; the processor receives a sentiment graph, each of which defines a sentiment; the processor generates a graph sentiment score based on the non-domain-specific knowledge graph and the sentiment graph; the processor generates a final sentiment score based on the graph sentiment score and the domain-specific machine learning sentiment score; and the processor determines the information sentiment in the digital information via the final sentiment score. In some aspects, the system may also include a method for automatically updating entity attributes in at least one database or server based on the final sentiment score.

[0012] In various aspects, an automated system includes instructions for: training a first machine learning model having a base layer and a second layer for the digital information; incorporating the base layer trained for the digital information into a second machine learning model; and training the second machine learning model including the base layer and the final layer to generate the domain-specific machine learning sentiment score. In other aspects, the instructions for training the first machine learning model include instructions for training to classify the subject matter of the digital information.

[0013] In various aspects, the automatic system includes instructions for generating the graph sentiment score, including: determining the graph similarity between each emotion graph in the emotion graph and the non-domain-specific knowledge graph; applying the emotions defined by each emotion graph in the emotion graph to its determined graph similarity to produce a graph-specific similarity tone score; and combining the graph-specific similarity tone scores of the emotion graphs.

[0014] In various aspects, the system includes instructions for generating the final sentiment score, including: applying a weighting to the graph sentiment score to generate a weighted graph sentiment score; applying another weighting to the domain-specific machine learning sentiment score to generate a weighted domain-specific machine learning sentiment score; and combining the weighted graph sentiment score and the weighted domain-specific machine learning sentiment score.

[0015] In various aspects, the element includes at least one of an entity, a name, a location, a time, or an event. In some aspects, at least a portion of the digital information is labeled. In some aspects, the emotions defined by each emotion map are related to a background context provided in a digital manner.

[0016] In various aspects, the present disclosure provides a non-transitory computer-readable storage medium having a program embodied thereon, the program being executable by a processor to perform a method for providing emotions of digital information, the method comprising: exporting the digital information from a source by the processor; generating a domain-specific machine learning emotion score by the processor based on the digital information through one of at least two machine learning models; autonomously drawing a non-domain-specific knowledge graph of associations between elements in a set of digital background information by the processor; receiving an emotion graph by the processor, each emotion graph defining an emotion; generating a graph emotion score by the processor based on the non-domain-specific knowledge graph and the emotion graph; generating a final emotion score by the processor based on the graph emotion score and the domain-specific machine learning emotion score; and determining the information emotion in the digital information by the processor via the final emotion score.

[0017] In various aspects, the program executable by the processor to perform the method further includes automatically updating the entity attributes in at least one of a database or a server based on the final sentiment score.

[0018] In various aspects, an automated system for updating stored entity attributes based on determined information sentiment is disclosed, the system comprising a database containing entity attributes, a processor, and a computer-readable medium, the computer-readable medium storing instructions, the instructions executable by the processor to perform the following operations: the processor inputs domain-specific digital information received from a source into a trained domain-specific machine learning model; the processor outputs a domain-specific sentiment score generated by the domain-specific machine learning model; the processor inputs digital news information into a knowledge graph representing an entity; the processor updates the knowledge graph with the digital news information; the processor determines the similarity between the knowledge graph and a defined sentiment graph to generate a graph sentiment score; the processor generates an entity sentiment score based on the domain-specific sentiment score and the graph sentiment score; the processor searches for an entity sentiment score entry stored in the database; and the processor automatically updates the entity sentiment score entry in the database based on the difference between the entity sentiment score and the entity sentiment score entry.

[0019] In various aspects, a connected system consisting of a node cluster to create and update an entity profile based on real-time information, the system comprising: a plurality of nodes connected within the cluster; a first node among the plurality of nodes in communication with a digital information channel, the first node comprising instructions executable to receive digital information associated with an entity from the digital information channel, input the digital information into a trained ML network to generate a sentiment classification, and output the sentiment classification to a processing node among the plurality of nodes; and a second node among the plurality of nodes in communication with a news source, the second node comprising instructions to receive digital news content from the news source and to draw a knowledge graph associated with the entity based on the digital news content. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In the description, for the purpose of explanation rather than limitation, specific details are set forth, such as particular aspects, procedures, techniques, etc., to provide a thorough understanding of the technology of the present invention. However, it will be apparent to those skilled in the art that the technology of the present invention can be practiced in other aspects different from these specific details.

[0021] The accompanying drawings, together with the detailed description below, are incorporated into and form a part of the specification and are used to further illustrate aspects of the concepts including the claimed disclosure and to explain various principles and advantages of those aspects. In the accompanying drawings, like reference numerals refer to the same or functionally similar elements throughout the different views.

[0022] The methods and systems disclosed herein have been represented by conventional symbols in the drawings where appropriate, showing only those specific details relevant to understanding the various aspects of the disclosure so as not to obscure the disclosure with details that would be apparent to one of ordinary skill in the art having the benefit of the description herein.

[0023] Figure 1 is a diagram of a method for determining sentiment in digital information according to at least one aspect of the present disclosure.

[0024] Figure 2 An entity graph representing a knowledge graph or a sentiment knowledge graph according to at least one aspect of the present disclosure is shown.

[0025] Figure 3 is a logic flow diagram of a method for determining sentiment in digital information in accordance with at least one aspect of the present disclosure.

[0026] Figure 4 is a logic flow diagram of a method for training a machine learning network according to at least one aspect of the present disclosure.

[0027] Figure 5 is a flow chart of a method for generating a sentiment score based on a drawn graph according to at least one aspect of the present disclosure.

[0028] Figure 6 is a block diagram of a computer device according to at least one aspect of the present disclosure.

[0029] Figure 7 is a diagrammatic representation of an example system including a host within which a set of instructions may be executed to perform any one or more of the methodologies discussed herein in accordance with at least one aspect of the present disclosure. DETAILED DESCRIPTION

[0030] Before discussing specific aspects and examples, some description of the terms used herein is provided below.

[0031] An "application" may include any software module configured to perform one or more specific functions when executed by a processor of a computer. For example, a "mobile application" may include a software module configured to be operated by a mobile device. An application may be configured to perform many different functions. For example, a "payment application" may include a software module configured to store and provide account credentials for transactions. A "wallet application" may include a software module with similar functionality to a payment application, which has multiple accounts that are pre-configured or registered so that they can be used through the wallet application. An "application" may be computer code or other data stored on a computer-readable medium (e.g., a memory element or a secure element) that can be executed by a processor to complete a task.

[0032] As used herein, the term "computing device" or "computer device" may refer to one or more electronic devices configured to communicate directly or indirectly with or on one or more networks. The computing device may be a mobile device, a desktop computer, or the like. As an example, the mobile device may include a cellular phone (e.g., a smartphone or a standard cellular phone), a portable computer, a wearable device (e.g., a watch, glasses, lenses, clothing, etc.), a personal digital assistant (PDA), and / or other similar devices. The computing device may be a mobile device or a non-mobile device, such as a desktop computer. In addition, the term "computer" may refer to any computing device that includes the necessary components for sending, receiving, processing, and / or outputting data and typically includes a display device, a processor, a memory, an input device, a network interface, and the like.

[0033] As used herein, the term "server" may include one or more computing devices, which may be individual stand-alone machines located in the same or different locations, may be owned or operated by the same or different entities, and may also be one or more clusters of distributed computers or "virtual" machines housed in a data center. It should be understood and appreciated by those skilled in the art that the functions performed by one "server" may be spread across multiple different computing devices for various reasons. As used herein, "server" is intended to refer to all such scenarios and should not be interpreted or limited to a specific configuration. In addition, the server as described herein may, but need not, reside at (or be operated by) an agent of a merchant, a payment network, a financial institution, a medical provider, a social media provider, a government agency, or any of the foregoing entities. The term "server" may also refer to or include one or more processors or computers, storage devices, or similar computer arrangements that operate or facilitate communication and processing by multiple parties in a network environment such as the Internet, but it should be understood that communication may be facilitated by one or more public or private network environments, and various other arrangements are possible. In addition, multiple computers (e.g., servers) or other computerized devices (e.g., point-of-sale devices) that communicate directly or indirectly in a network environment may constitute a "system" (e.g., a merchant's point-of-sale system). As used herein, references to a "server" or "processor" may refer to a previously stated server and / or processor stated as performing a previously stated step or function, different servers and / or processors, and / or combinations of servers and / or processors. The term processor may also mean that the methods disclosed herein may be practiced by distributed processors all under the control of a payment network server.

[0034] As used herein, the term "system" may refer to one or more computing devices or a combination of computing devices (eg, processors, servers, client devices, software applications, components of such computing devices, etc.).

[0035] The terms "client device" and "user device" refer to any electronic device configured to communicate with one or more servers or remote devices and / or systems. Client devices or user devices may include mobile devices, network-enabled devices (e.g., network-enabled televisions, refrigerators, thermostats, and / or the like), computers or computing devices, POS systems, and / or any other devices or systems capable of communicating with a network. Client devices may also include desktop computers, laptop computers, mobile computers (e.g., smart phones), wearable computers (e.g., watches, glasses, lenses, clothing, etc.), cellular phones, network-enabled appliances (e.g., network-enabled televisions, refrigerators, thermostats, etc.), point of sale (POS) systems, and / or any other devices, systems, and / or software applications configured to communicate with a remote device or system. "Users" may include individuals. In some aspects, users may be associated with one or more personal accounts and / or mobile devices. Users may also be referred to as cardholders, account holders, or consumers.

[0036] Information, current events, data, and news clips are all factors in assessing the creditworthiness, credit risk, and other status of individuals and entities when building accurate ratings, entity, and risk profiles so that automated decisions or updates can be made to systems and / or databases. For example, in the savings, loan, or credit industry, profiles constructed for individuals, organizations, and other entities are used to design or provide customizable products to these entities. Operations teams routinely search for information and news articles about companies or businesses to determine their credit worthiness and risk, however, there is no objective view or understanding of what constitutes a negative or positive indicator, tone, or sentiment about an entity in one or more related areas, nor is there a way to scale or automate the process using current technology in the field.

[0037] One approach to scale these processes is to attempt to automate the process using a trained machine learning model that is not only able to determine the category and domain of information, news items, and other data, but also able to classify the sentiment associated with each contextual information, news item, or entity in the information provided. However, it is difficult to automate such processes or train a network model to determine the sentiment of contextual information associated with one or more entities. Some of the challenges faced when training autonomous models include the lack of available data, the lack of well-labeled information and data, the difficulty in classifying information and domains based on this limited information and data, errors and biases in the available data, and the significant cost and investment associated with obtaining well-labeled data of consistent quality.

[0038] For example, a model trained on general domain labels such as eBay labeled data, or derived from customer reviews from merchant websites such as Amazon, Delta airlines, or other similar publicly available data, may contain sentiments that can be identified, but those sentiments may not be relevant to the use case, domain, or subject matter that is desired and used. For example, a negative customer review of Delta's seating arrangement or in-flight entertainment may capture general customer sentiment about the ride, but it is likely not relevant to Delta Airlines' credit worthiness or direct bankruptcy risk.

[0039] Another option to improve data relevance is to filter the data or available information based on keywords, with the goal of this filtering or keyword approach being to select data that may be more relevant to a particular use case or network training model. However, filtering may be an overly direct approach and is not a well-regulated option because keywords or tags may be too broad or too limited in approach, and the selection or use of selected specific words or phrases may include inherent biases (and the failure to select or use other words and phrases may also include inherent biases that will affect the data). A more refined approach that takes into account a more complete perspective of contextual information is needed to accurately assess sentiment about entities in a specific domain.

[0040] Systems and methods are presented herein to automatically determine sentiment in various news and digital information articles by training, improving, and enhancing machine learning networks with non-domain-specific entity graphs and using sentiment determination to update entity attributes, databases, and make automated changes and decisions based on constructed entity profiles and sentiment information. The goal of the presented systems and methods is to develop automatic, accurate, and efficient processes for different use cases to extract, derive, find, and learn sentiment in background information in various domains. The presented techniques include two main parts, the first of which is to train machine learning models using auxiliary information to learn how to classify domains and then learn how to classify specific sentiments with partially labeled and freely available data; and the second part is to enhance these models with specially tuned non-domain-specific knowledge and sentiment graphs.

[0041] Although there are different machine learning techniques for understanding language, such as various natural language processing techniques, machine learning models have not always been able to accurately derive emotions from input information. The present disclosure provides techniques for more accurately deriving emotions from input information such as, but not limited to, digital text information, via machine learning models, as well as techniques for training such machine learning models to improve their ability, reliability, and accuracy in deriving emotions from information input.

[0042] In various aspects of the present disclosure, improved sentiment derivation techniques allow the machine learning models presented herein to be incorporated into systems such as enterprise systems or networks (these are referred to herein as "enterprises") to automatically and reliably update information stored in a database or data warehouse, for example, based on the improved sentiment understanding output of the machine learning models. According to the models proposed herein, automatic updating of data based on these machine learning models can only be performed if the machine learning models can be trusted to accurately derive sentiment information from various information sources. For example, real-time export of sentiment information about individuals and / or entities can cause updates to sentiment levels / scores stored in an enterprise database, which in turn can affect various results, such as the credit or risk rating of an entity, and therefore affect the permissions and behavior of the entire system.

[0043] As a non-limiting example, real-time news information about a coup in a country may cause the machine learning model proposed herein to detect negative sentiment associated with the country, which may then be associated with entities doing business in the country or with the country or its related entities, or may put a certain amount of business activity at risk, thereby increasing its designated risk score. The enterprise system may use these new scores to update the entity / individual credit rating or reputation, and based on pre-configured or predetermined thresholds, may automatically prevent the approval of loans or certain sizes of loans through an automatic override mechanism on the enterprise network. Of course, the opposite is also incorporated into the system described herein, and when positive sentiment is associated with an entity, such as a successful IPO or product launch of a company, the data, risk scores and permissions of the entire enterprise may be automatically updated to allow new permissions, such as a larger loan to be granted / additional stock to be purchased, or additional / increased credit lines to be provided by the enterprise network or users of the enterprise network / system to be issued to related companies / entities / individuals.

[0044] In various embodiments, sentiment scores may be used to provide actionable alert notifications on a user interface of a system, network, or enterprise user / administrator (referred to herein as a "user"). In one non-limiting example, a new or updated score, such as a risk or credit score based on receiving, processing, and deriving new or updated sentiment about an individual / entity and the sentiment associated with the individual / entity. For example, a notification may indicate an alert of a sudden excess of new sentiment about a particular entity or individual (e.g., a large amount of good sentiment following a successful product launch for an entity). The user may be provided with the option to further drill down into the information behind the sentiment or take other action against the individual / entity.

[0045] Additionally, actionable notifications may be displayed to certain users, such as those who are set / permitted to receive such notifications and / or who are able to approve or deny updates, such as scores or attributes in a database, or approve / deny actions, such as increasing a credit line or issuing funds or loans to a person / entity. Notifications may be provided to users, such as on an interactive graphical user interface on a device of the system or enterprise, where the notifications may include any of the following: sentiment information, reasons why sentiment information is classified in a certain manner, scores affected by sentiment information, ratings affected by updated scores and / or sentiment information, selectable actionable options to take actions, such as: allowing database updates, accepting, rejecting incoming sentiment information, scores, or ratings, allowing new / updated sentiment information, scores, or ratings to be incorporated into or applied / implemented into actions, decisions, rules, or permissions associated with a particular person or entity, or adjusting any of the above to increase or decrease, for example, the scope of permissions available to a person / entity.

[0046] While the present technology is susceptible to aspects in many different forms, several specific aspects are shown in the drawings and will be described in detail herein, it being understood that this disclosure is to be considered as an exemplification of the principles of the present technology and is not intended to limit the technology to the aspects shown.

[0047] Figure 11 is a diagram of a method 100 for determining emotions in digital information according to at least one aspect of the present disclosure. The method 100 for determining emotions in digital information is based on a combination of two separate parts. The first part of the method 100 includes training a machine learning network. In one aspect, the first part includes two machine learning networks: a first machine learning network 105 (also referred to herein as "M1") and a second machine learning network 110 (also referred to herein as "M2"). M1105 defines two separate neural layers: a base layer 115 and a final layer 120 (also referred to herein as "L1"). M1105 is trained to provide a general subject or domain classification 125 as an output from data input fed into both the base layer 115 and the final layer L1120. Once sufficient training has occurred, the base layer 115 is incorporated into M2110, which is in turn trained to provide a domain-specific emotion classification 140 having a combination of the trained base layer 115 and the final layer L2135. The trained base layer 115 and the final layer L2135 together constitute M2110. The sentiment classification output by M2110 is represented by a vector or classification score Ys 145. In several aspects of the present disclosure, both M1105 and M2110 use the same input or input from the same or similar input pool, however the goals or target outputs of M1105 and M2110 are different. In various aspects, for example, M1105 has a goal of subject classification or categorization of the input, while M2110 has a goal of sentiment classification of the input. This technique allows M1105 and its base layers to be trained for easier tasks and goals, i.e., classifying subjects as freely available digital information, and then using the already trained base layers as part of M2110 to produce different outputs that require more advanced training, i.e., domain-specific sentiment classification.

[0048] The second part of the method 100 includes determining non-domain-specific emotions in digital information. The second part of the method 100 regards emotion classification as a graph matching task and includes extracting information or news from at least one source 150. In one aspect, this information is, for example, digital information. The source 150 can be an information source, a database, a news source, and the information can be formatted as audio, video, image, text, or any other suitable digital format. In various aspects, the information extracted, derived, or received from the at least one source 150 can be news or current event information. This information is used to construct 155 a non-domain-specific knowledge graph 160, which includes various components of the derived information or news. For example, the knowledge graph 160 can represent or build on the relationship between entities, events, places, names, dates, emotions, topics, etc., any of which can be represented on the knowledge graph 160, depending on what associations the knowledge graph 160 intends to present. The knowledge graph 160 can represent or include one or more news information fragments or information from various news sources 150. In various aspects, the digital information received or extracted from news source 150 relates to an entity, such as a company, individual, institution, state, government, etc.

[0049] In various aspects, users 165, including domain users, may define 170 emotions for various news segments, information, or current events or other forms of digital information. The system then uses these defined emotions to create or construct an emotion knowledge graph 175 (also referred to herein as a "emotion graph"), which includes various graphs representing different emotion types. The system may refer to one or more computing devices or a combination of computing devices (e.g., a processor, a server, a client device, a software application, components of these computing devices, and / or the like).

[0050] Exemplary emotion types may include, but are not limited to, positive, negative, neutral, slightly negative, moderately negative, highly negative, slightly positive, moderately positive, and highly positive. The knowledge graph 160 is then compared to various emotion knowledge graphs 175 via a graph similarity function 180g(Vec(E), Vec(Ii)), where Vec(E) represents a vector or value of the knowledge graph 160, and Vec(Ii) represents a vector or value (or vectorized value) of the emotion graph 175 compared to the knowledge graph 160. The graph similarity function may be weighted or multiplied by the defined emotions extracted from or attributed to the relevant emotion graph 175 to produce a graph emotion score for the particular emotion graph 175. Each graph similarity function 180 is weighted using the defined emotions of the relevant emotion knowledge graph 175. The graph similarity function 180 is then combined 190 with other graph emotion scores to derive a total graph emotion score Yg 192. The overall graph sentiment score Yg 192 is then combined with the machine learning sentiment score Ys 145 to produce a final sentiment score Y 195 representing the sentiment of the entity. Each of the overall graph sentiment score Yg 192 and the classification score Ys 145 may also be weighted when combined to produce the final sentiment score 195.

[0051] Figure 2 An entity graph 200 representing a knowledge graph or a sentiment knowledge graph according to at least one aspect of the present disclosure is shown. The entity graph 200 includes a number of nodes 204. Each of the nodes 204 corresponds to an attribute from a set of attributes (e.g., including attribute 1 2021, attribute 2 2022, attribute 3 2023, attribute 4 2024, attribute 5 2025, attribute 6 2026, attribute 7 2027, attribute 8 2028, ... and attribute 202 n Attribute 202 of an attribute set (e.g., an attribute set of an entity graph). Entity graph 200 may define news or information that may be related to an entity, an unknown entity graph, or a known entity graph, as well as sentiment defined by a user or reader of the information. Thus, attribute 202 may correspond to an entity, an entity attribute, a news story, a name, a location, a sentiment, a location, a date, an event, and other news information extracted from an information source, a news source, or other database. In addition, any number (e.g., any positive integer) of attributes may be included in the attribute set. Accordingly, entity graph 200 may include any number (e.g., any positive integer) of nodes 204.

[0052] Still refer to Figure 2, the structure of the entity graph 200 is determined by the placement of edges 206. In general, each node 204 is connected to at least one other node 204 via an edge 206. Some nodes 204 may be connected to multiple other nodes 204 via multiple edges. For example, a node 204 corresponding to attribute 2 2022 is only connected to a node 204 corresponding to attribute 1 2021 via an edge 206. Conversely, a node 204 corresponding to attribute 1 2021 is connected to nodes 204 corresponding to attributes 2 2022, 3 2023, 4 2024, 7 2027, 8 2028, and 9 2029 via edges 206. n 202 n Node 204. Although Figure 2 One particular entity graph structure is depicted in the non-limiting aspects of FIG. 2 , but the entity graph 200 may have any structure in which each node 204 is connected to one or more of any other nodes 204 via edges 206 .

[0053] Figure 3 is a logic flow diagram of a method 300 for determining sentiment in digital information according to at least one aspect of the present disclosure. According to the method 300, information is first derived 305 from a source. The source may be a news source, an information source, or any other background information source, and in various aspects, the source is a digital information source. This information is then used by one or more processors to generate 310 a domain-specific machine learning (also referred to herein as "ML") sentiment score based on the digital information through one of at least two machine learning models. The two models are Figure 1 The training of the machine learning model and the generation of domain-specific sentiment scores are described in detail in Figure 4 middle.

[0054] Still refer to Figure 3 , domain-specific ML scores are provided by the first part of the method 300, while the second part of the method 300 treats sentiment classification as a graph matching task. The method 300 then autonomously or automatically draws 315, generates or constructs at least one knowledge graph, which describes how various entities are related, associated or interact with each other. This drawing 315 can be performed based on exporting 305 information or from a separate information extraction or export step (not shown) and / or from a separate information source. The information source can be an internal or external (e.g., third-party) database, server, or system's digital information source. The knowledge graph is automatically drawn 315 by the system to show various relationships between several entities in one or more graphs. The type of relationship or entity can be customized in the system. The drawing 315 of the knowledge graph can be used to find or identify one or more entities from a digital information source.

[0055] In some aspects, a knowledge graph is constructed to illustrate entity interactions and associations, while in other aspects, a knowledge graph is created through autonomous drawing 315 to illustrate associations between various different elements, including but not limited to entities, events, places, names, dates, or any other relevant information derived from a news source. The system, database, server, or other processor may then receive 320 other sentiment graphs generated based on information, digital information, or feedback provided by domain users, whereby the user defines or determines the emotions in various different articles, information fragments, news information, or background information fragments. In various aspects, the graph is generated internally and uses a source separate from the source used to autonomously draw 315 the knowledge graph. These sentiment graphs are used to represent different emotion scores or emotional states, and may correspond to, for example, positive emotions or tones, negative emotions, or neutral emotions. By comparing the autonomously drawn 315 knowledge graph with the received 320 sentiment graph via, for example Figure 1 The graph similarity function 180 shown in is compared with other graph similarity functions to generate 325 a non-domain specific graph sentiment score.

[0056] In some aspects, similarities between graphs are calculated based on graph embeddings, and then the embedded vector similarities are compared to distance measurements. The system then automatically generates 330 a final sentiment score by combining domain-specific machine learning sentiment scores and non-domain-specific graph sentiment scores. Weights may be added to any of these scores, which weights may be the product of machine learning network training, or may be assigned based on other factors, observations, heuristics, or other settings. The sentiment score that is finally generated may be used to determine the sentiment of the information source on which the derived or extracted digital information or knowledge and / or sentiment graph is based. In various aspects, the sentiment score that is finally generated is related to a specific entity that may be mentioned in the digital information.

[0057] Figure 4 is a logic flow diagram of a method 400 for training a machine learning network according to at least one aspect of the present disclosure. The method 400 details one aspect, wherein the training Figure 3 The two machine learning models described in Figure 3 The method 300 shown in step 310 generates a domain-specific machine learning sentiment score. Now referring to Figure 3 and 4 , extract 405 or export 305 information from a source as at least partially labeled background information. The exported 305 ( Figure 3 ) or extract 405( Figure 4), and the information may be in various formats, including audio, video, images, text, or other suitable digital formats. In some aspects of the present disclosure, the information extracted, derived, or received from the information source may be general background information including news or current event information. This information has been labeled or populated with categories or classified in one or more ways. For example, this information may be labeled with a label such as "stock price". The extraction of this information may include extracting background information that may, for example, describe the subject of the news article, and extracting other information that is generally available and provided by the news source and is at least partially labeled with labels or categories.

[0058] Reference now Figure 1 and 4 , because the information is at least partially labeled, the information extracted or derived 405 can be used as input to a basic machine learning model to perform a task in, for example, M1105 ( Figure 1 ) without any additional domain specific knowledge. In various aspects, this partially labeled and publicly available information is used to train 410 a first machine learning model corresponding to M1105. The first machine learning model may include two layers, namely a base layer and a final layer, and is trained 410 using a large amount of labeled data to determine the subject matter or classification of the information derived or extracted in 405.

[0059] Still reference Figure 1 and 4 , in model M1105( Figure 1 ) is trained 410, the trained base layers of the model are copied or incorporated 415 into the corresponding M2110 ( Figure 1 ). Because the base layers have been fine-tuned in M1105 to classify topics or categories, or trained to provide outputs about any other suitable goal based on training on a large amount of at least partially labeled data, model M2110, especially its final layer L2135 ( Figure 1 ), can be adjusted to train 420 only to determine, output or generate 425 domain, category or topic specific sentiment or sentiment scores. For example, this second model M2110 can generate an output score, value or vector indicating whether an article or other piece of information is negative, positive or neutral on a category or topic that the model has identified. This output is a domain specific machine learning sentiment score or classification Ys145 ( Figure 1 ). Finally, in various aspects, determining the weights to apply to the output may be learned during training of M1105 or M2110, and may also be output by M2110. These weights may be applied 430 to the generated 425 domain-specific machine-learned sentiment scores.

[0060] Figure 5is a logic flow diagram of a method 500 for generating a non-domain-specific emotion score based on a drawn graph according to at least one aspect of the present disclosure. According to the method 500, digital information input is first received 505 by a computing device, server, or system. This digital information can be compared with Figure 3 and 4 The digital information marked as 305 and 405 in the figure are the same, or come from the same source, or can be other digital information received separately from other sources. The system then autonomously draws 510 a non-domain-specific knowledge graph of associations between elements in a set of digital background information through a processor, server, or computing device. This autonomous drawing 510 provides a knowledge graph of associations between different elements identified in the digital information. The elements can be entities, wherein the drawing associates the entities with each other, and in other aspects, the elements can be various other features or elements, such as names, places, dates, events, entities, and the knowledge autonomous drawing 510 connects all of these various features to each other.

[0061] According to the method 500, the processor emotion map or emotion knowledge map receives 520 each emotion map that defines the emotion or tone of the information. These can be emotion maps generated autonomously by a computing device, server or system, or the maps can be generated by a domain-specific user who assigns emotion scores or emotion attributes to each digital information. The various emotion scores, attributes and / or emotions / tones provided can then be autonomously mapped 510 into an emotion map that represents the tone or emotion of the digital information. Figure 2 An example of the structure of a knowledge graph or sentiment graph is shown (see entity graph 200). Examples include receiving sentiment scores or sentiment descriptions from domain users, such as about various situations and / or news or current projects. Sentiments can be negative, positive, neutral, or between these parameters.

[0062] Another example may include a domain user or an autonomous system generating a score for each information item or digital information, and the sentiment score associated with each item is then autonomously plotted onto a graph, wherein the graph represents an average sentiment score, which in turn may represent positive sentiment, negative sentiment, neutral sentiment (or sentiment between these categories). A graph similarity function may be applied to determine 530 the similarity between the knowledge graph plotted and various sentiment graphs via a graph similarity function, which may be, for example, Figure 1The graph similarity function 180g (Vec (E), Vec (Ii)) is shown. For each graph similarity function comparing a particular emotion graph to a knowledge graph, the emotion score of the particular emotion graph is applied 540. The emotion score applied 540 can be multiplied by the graph similarity function to produce a product between each related emotion graph-knowledge graph combination. The product of each emotion score and graph similarity function is combined with the products of all other emotion scores and graph similarity functions, e.g. Figure 1 As seen in equation 190, to generate 550 Figure 1 Graph-specific similarity tone scores of the emotional graph Yg192 in .

[0063] Figure 6 1 is a block diagram of a computer device 1000 according to at least one aspect of the present disclosure. The computer device 1000 or its subsystems may be used to perform the methods and functions described herein. The example computer device 1000, also referred to herein as the subsystem 1000, is interconnected via a system bus 1010. Additional subsystems are shown, such as a printer 1018, a keyboard 1026, a fixed disk 1028 (or other memory including a computer-readable medium), a display screen 1022 coupled to a display adapter 1020, and the like. Peripheral devices and I / O devices coupled to an input / output (I / O) controller 1012 (which may be a processor or other suitable controller) may be connected to the computer system by any number of components known in the art, such as a serial port 1024. For example, the serial port 1024 or the external interface 1030 may be used to connect the computer device to a wide area network such as the Internet, a mouse input device, or a scanner. The interconnection via system bus 1010 allows central processor 1016 to communicate with each subsystem and control the execution of instructions from system memory 1014 or fixed disk 1028 and the exchange of information between subsystems. System memory 1014 and / or fixed disk 1028 may embody computer-readable media.

[0064] Figure 72000 is a diagrammatic representation of an example system 1 including a host 2000 in accordance with at least one aspect of the present disclosure, within which an instruction set for performing any one or more of the methods discussed herein may be executed. In various aspects, the host 2000 operates as a standalone device, or may be connected (e.g., networked) to other machines. In a networked deployment, the host 2000 may operate in the capacity of a server or client machine in a server-client network environment, or as a peer machine in a peer-to-peer (or distributed) network environment. The host 2000 may be a computer or computing device, a personal computer (PC), a tablet PC, a set-top box (STB), a personal digital assistant (PDA), a cellular phone, a portable music player (e.g., a portable hard drive audio device, such as a Moving Picture Experts Group Audio Layer 3 (MP3) player), a network appliance, a network router, a switch or a bridge, or any machine capable of executing an instruction set (sequentially or otherwise) that specifies an action to be taken by the machine. Further, while a single machine is illustrated, the term "machine" shall also be taken to include any collection of machines that individually or jointly execute a set (or multiple sets) of instructions to perform any one or more of the methodologies discussed herein.

[0065] The example system 1 includes a host 2000 running a host operating system (OS) 2001 on a processor or multiple processors / processor cores 2003 (e.g., a central processing unit (CPU), a graphics processing unit (GPU), or both) and various memory nodes 2005. The host OS 2001 may include a hypervisor 2004 that is capable of controlling functions and / or communicating with a virtual machine (“VM”) 2010 running on a machine-readable medium. The VM 2010 may also include a virtual CPU or vCPU 2009. Memory nodes 2005 and 2007 may be linked or pinned to a virtual memory node or vNode 2006, respectively. When a memory node 2005 is linked or pinned to a corresponding virtual node 2006, data may then be mapped directly from the memory node 2005 to its corresponding vNode 2006.

[0066] All the various components shown in the host 2000 can be connected to each other and connected to each other, or communicate with each other via a bus (not shown) or via other coupling or communication channels or mechanisms. The host 2000 may also include a video display, an audio device or other peripheral device 2020 (e.g., a liquid crystal display (LCD), an alphanumeric input device including, for example, a keyboard, a cursor control device (e.g., a mouse, a voice recognition or biometric verification unit, an external drive, a signal generating device, such as a speaker), a permanent storage device 2002 (also referred to as a disk drive unit) and a network interface device 2025. The host 2000 may also include a data encryption module (not shown) for encrypting data. The components provided in the host 2000 are those components commonly found in computer systems that are suitable for use with various aspects of the present disclosure, and are intended to represent a wide range of such computer components known in the art. Therefore, the system 1 can be a server, a minicomputer, a mainframe computer, or any other computer system. The computer may also include different bus configurations, network platforms, multi-processor platforms, etc. Various operating systems may be used, including UNIX, LINUX, WINDOWS, QNXANDROID, IOS, CHROME, TIZEN, and other suitable operating systems.

[0067] The disk drive unit 2002 may also be a solid state drive (SSD), a hard disk drive (HDD), e.MMC, and UFS or other computer or machine readable media, on which one or more instruction sets and data structures (e.g., data or instructions 2015) embodying or utilizing any one or more methods or functions described herein are stored. The instructions 2015 may also reside completely or at least partially in the main memory node 2005 and / or in the processor 2003 during the execution of the instructions by the host 2000. The processor 2003 and the memory node 2005 may also include machine readable media.

[0068] The instructions 2015 may also be sent or received via the network 2030 via the network interface device 2025 using any of several well-known transmission protocols (e.g., Hypertext Transfer Protocol (HTTP)). The term "computer-readable medium" or "machine-readable medium" should be understood to include a single medium or multiple media (e.g., centralized or distributed databases and / or associated cache memories and servers) that store one or more sets of instructions. The term "computer-readable medium" should also be considered to include any medium that can store, encode, or carry an instruction set for execution by the machine 2000 and cause the machine 2000 to perform any one or more methods of the present application, or can store, encode, or carry a data structure utilized by or associated with such an instruction set. Therefore, the term "computer-readable medium" should be understood to include, but is not limited to, solid-state memory, optical and magnetic media, and carrier signals. This medium may also include, but is not limited to, hard disks, floppy disks, flash memory cards, digital video disks, random access memories (RAM), read-only memories (ROM), etc. The exemplary aspects described herein may be implemented in an operating environment that includes software installed on a computer, installed in hardware, or a combination of software and hardware.

[0069] Those skilled in the art will recognize that an Internet service can be configured to provide Internet access to one or more computing devices coupled to the Internet service, and that the computing devices may include one or more processors, buses, memory devices, display devices, input / output devices, etc. In addition, those skilled in the art will appreciate that the Internet service can be coupled to one or more databases, repositories, servers, etc., which can be used to implement any of the various aspects of the present disclosure described herein.

[0070] The computer program instructions may also be loaded onto a computer, server, other programmable data processing device or other apparatus so that a series of operating steps are performed on the computer, other programmable device or other apparatus to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide a process for implementing the functions / actions specified in one or more boxes of the flowchart and / or block diagram.

[0071] For example, a suitable network may include or interface with any one or more of the following: a local intranet, a PAN (personal local area network), a LAN (local area network), a WAN (wide area network), a MAN (metropolitan area network), a virtual private network (VPN), a storage area network (SAN), a frame relay connection, an advanced intelligent network (AIN) connection, a synchronous optical network (SONET) connection, a digital T1, T3, E1 or E3 line, a digital data service (DDS) connection, a DSL (digital subscriber line) connection, an Ethernet connection, an ISDN (integrated services digital network) line, a dial-up port (e.g., a V.90, V.34 or V.34bis analog modem connection), a cable modem, an ATM (asynchronous transfer mode) connection, or an FDDI (fiber distributed data interface) or CDDI (copper distributed data interface) connection. In addition, communications may include links to any of a variety of wireless networks, including WAP (Wireless Application Protocol), GPRS (General Packet Radio Service), GSM (Global System for Mobile Communications), CDMA (Code Division Multiple Access) or TDMA (Time Division Multiple Access), cellular telephone networks, GPS (Global Positioning System), CDPD (Cellular Digital Packet Data), RIM (Research In Motion Ltd.) duplex paging network, Bluetooth radio, or IEEE 802.11 based radio frequency networks. Network 3030 may also include RS-232 serial connections, IEEE-1394 (FireWire) connections, Fibre Channel connections, IrDA (infrared) ports, SCSI (Small Computer System Interface) connections, USB (Universal Serial Bus) connections, or other wired or wireless, digital or analog interfaces or connections, mesh or Any one or more of the network or interface therewith.

[0072] Generally speaking, a cloud-based computing environment is a resource that typically combines the computing power of large groups of processors (e.g., within a web server) and / or the storage capacity of large groups of computer memory or storage devices. Systems that provide cloud-based resources may be employed solely by their owners, or such systems may be accessed by external users who deploy applications within the computing infrastructure to gain the benefits of large computing or storage resources.

[0073] The cloud is formed, for example, by a network of network servers including multiple computing devices, such as host 2000, where each server 2035 (or at least multiple servers) provides processor and / or storage resources. These servers manage workloads provided by multiple users (e.g., cloud resource customers or other users). Typically, each user's workload demand for the cloud changes in real time, sometimes even greatly. The nature and extent of these changes usually depends on the type of business associated with the user.

[0074] It is noteworthy that any hardware platform suitable for performing the processing described herein is suitable for use with the technology. As used herein, the terms "computer-readable storage medium" and "computer-readable storage medium" refer to any one or more media that participate in providing instructions to the CPU for execution. This medium can take a variety of forms, including (but not limited to) non-volatile media, volatile media, and transmission media. Non-volatile media include (for example) optical or magnetic disks, such as fixed disks. Volatile media include dynamic memory, such as system RAM. Transmission media include coaxial cables, copper wires, and optical fibers, among others, including wires containing one aspect of a bus. Transmission media can also take the form of sound waves or light waves, such as those generated during radio frequency (RF) and infrared (IR) data communications. Common forms of computer readable media include, for example, floppy disks, hard disks, magnetic tape, any other magnetic media, CD-ROMs, digital video disks (DVDs), any other optical media, any other physical media with markings or patterns of holes, RAM, PROM, EPROM, EEPROM, FLASH EPROM, any other memory chip or data exchange adapter, carrier wave, or any other medium from which a computer can read.

[0075] Various forms of computer readable media may be involved in carrying one or more sequences of one or more instructions to the CPU for execution. The bus carries the data to the system RAM, from which the CPU retrieves the instructions and executes them. The instructions received by the system RAM may optionally be stored on a fixed disk before or after execution by the CPU.

[0076] Computer program code for performing operations on aspects of the present technology may be written in any combination of one or more programming languages, including object-oriented programming languages, such as Java, Smalltalk, C++, or the like, and conventional procedural programming languages, such as the "C" programming language, Go, Python, or other programming languages, including assembly languages. The program code may be executed entirely on the user's computer, partially on the user's computer; as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or an external computer may be connected (e.g., using an Internet service provider to connect via the Internet).

[0077] Examples of methods according to various aspects of the present disclosure are provided below in the following numbered clauses. One aspect of the method may include any one or more than one and any combination of the numbered clauses described below.

[0078] Clause 1. An automated computer-implemented method for determining information sentiment in digital information, the method comprising: a processor deriving the digital information from a source; the processor generating a domain-specific machine learning sentiment score based on the digital information through one of at least two machine learning models; the processor autonomously drawing a non-domain-specific knowledge graph of associations between elements in a set of digital background information; the processor receiving a sentiment graph, each sentiment graph defining a sentiment; the processor generating a graph sentiment score based on the non-domain-specific knowledge graph and the sentiment graph; the processor generating a final sentiment score based on the graph sentiment score and the domain-specific machine learning sentiment score; and the processor determining the information sentiment in the digital information via the final sentiment score.

[0079] Clause 2. The method of clause 1, further comprising automatically updating, by the processor, entity attributes in at least one of a database or a server based on the final sentiment score.

[0080] Clause 3. The method according to any one of clauses 1-2 further includes: the processor training a first machine learning model having a base layer and a second layer for the digital information; the processor incorporating the base layer trained for the digital information into the second machine learning model; and the processor training the second machine learning model including the base layer and the final layer to generate the domain-specific machine learning sentiment score.

[0081] Clause 4. A method according to any one of clauses 1-3, wherein the training of the first machine learning model includes training the first machine learning model to classify the subject matter of the digital information.

[0082] Clause 5. A method according to any one of clauses 1-4, wherein the generation of the graph emotion score includes: determining, by the processor, the graph similarity between each emotion graph in the emotion graph and the non-domain-specific knowledge graph; applying, by the processor, the emotions defined by each emotion graph in the emotion graph to the determined graph similarity to generate a graph-specific similarity tone score; and combining, by the processor, the graph-specific similarity tone scores of the emotion graphs.

[0083] Clause 6. A method according to any one of clauses 1-5, wherein the generation of the final sentiment score includes: the processor applying a weighting to the graph sentiment score to generate a weighted graph sentiment score; the processor applying another weighting to the domain-specific machine learning sentiment score to generate a weighted domain-specific machine learning sentiment score; and the processor combining the weighted graph sentiment score and the weighted domain-specific machine learning sentiment score.

[0084] Clause 7. The method of any of clauses 1-6, wherein the element comprises at least one of an entity, a name, a location, a time, or an event.

[0085] Clause 8. The method of any one of clauses 1-7, wherein at least a portion of the digital information is marked.

[0086] Clause 9. A method according to any one of clauses 1-8, wherein the emotions defined by each of the emotion graphs relate to a digitally provided background context.

[0087] Clause 10. An automated system for updating stored entity attributes based on determined information sentiment, the system comprising a database containing entity attributes, a processor, and a computer-readable medium, the computer-readable medium storing instructions, the instructions being executable by the processor to perform the following operations: the processor inputs domain-specific digital information received from a source into a trained domain-specific machine learning model; the processor outputs a domain-specific sentiment score generated by the domain-specific machine learning model; the processor inputs digital news information into a knowledge graph representing an entity; the processor updates the knowledge graph with the digital news information; the processor determines the similarity between the knowledge graph and a defined sentiment graph to generate a graph sentiment score; the processor generates an entity sentiment score based on the domain-specific sentiment score and the graph sentiment score; the processor searches for an entity sentiment score entry stored in the database; and the processor automatically updates the entity sentiment score entry in the database based on the difference between the entity sentiment score and the entity sentiment score entry.

[0088] Clause 11. A system according to Clause 10, wherein the automatic updating of the entity sentiment score entry includes at least one of: deleting, changing, adding to, subtracting from, or applying a weight to the entity sentiment score entry in the database.

[0089] Item 12. A connected system consisting of a node cluster for creating and updating entity profiles based on real-time information, the system comprising: a plurality of nodes connected within the cluster; a first node among the plurality of nodes that communicates with a digital information channel, the first node comprising instructions executable to receive digital information associated with an entity from the digital information channel, input the digital information into a trained ML network to generate a sentiment classification, and output the sentiment classification to a processing node among the plurality of nodes; and a second node among the plurality of nodes that communicates with a news source, the second node comprising instructions to receive digital news content from the news source and to draw a knowledge graph associated with the entity based on the digital news content.

[0090] Clause 13. The system of clause 12, wherein the second node is in further communication with a user server to receive a sentiment classification of content from the user server, wherein the sentiment classification is generated by a domain user.

[0091] Clause 14. The system of any of Clauses 12-13, wherein the user server includes instructions to receive new emotion classifications from domain users; and update a user emotion database storing the emotion classifications with the new emotion classifications received from the domain users.

[0092] Clause 15. The system according to any one of clauses 12-14, wherein the second node includes further instructions to receive the emotion classification from a domain user server; and draw an emotion map based on the emotion classification, wherein each emotion map contains an emotion tone.

[0093] Clause 16. A system according to any one of clauses 12-15, wherein the second node includes additional instructions to generate an entity sentiment score based on the similarity between the knowledge graph and at least one of the sentiment graphs, and output the entity sentiment score to the processing node.

[0094] Clause 17. A system according to any one of clauses 12-16, wherein the processing node includes instructions to perform the following operations: receiving the sentiment classification from the first node; receiving the entity sentiment score from the second node; determining a final entity sentiment score based on the sentiment classification and the entity sentiment score; and pushing the final entity sentiment score to a database node among a plurality of connected nodes, the database node storing a profile of the entity.

[0095] Clause 18. The system of any one of clauses 12-17, wherein the database node comprises instructions to: receive the final entity sentiment score from the processing node; and update a profile of the entity stored in the database node with the final entity sentiment score.

[0096] Clause 19. The system of any of clauses 12-18, wherein the first node includes instructions to detect the digital information from the digital information channel.

[0097] Clause 20. The system of any of clauses 12-19, wherein the second node includes instructions to detect the digital news content from the news source.

[0098] The foregoing detailed description includes references to the accompanying drawings, which form a part of the detailed description. The accompanying drawings show diagrams according to exemplary aspects. These exemplary aspects, also referred to herein as "examples," are described in sufficient detail to enable those skilled in the art to practice the inventive subject matter.

[0099] The various aspects described above are presented as examples only and not as limitations. The description is not intended to limit the scope of the present invention to the form set forth herein. On the contrary, as will be appreciated by those of ordinary skill in the art, this specification is intended to cover such alternatives, modifications and equivalents that may be included within the scope of the present invention.

[0100] Although specific aspects and examples of the system are described above for illustrative purposes, various equivalent modifications may be made within the scope of the system as will be appreciated by those skilled in the relevant art. For example, although processes or steps are presented in a given order, alternative aspects may execute routines with different orders, and some processes or steps may be deleted, moved, added, subdivided, combined, and / or modified to provide alternatives or sub-combinations. Each of these processes or steps may be implemented in a variety of different ways. In addition, although processes or steps are sometimes shown as being performed serially, these processes or steps may instead be performed in parallel, or may be performed at different times.

[0101] Without departing from the scope of the claimed protection, various aspects can be combined, other aspects can be utilized, or structural, logical and electrical changes can be performed. It should be understood by those skilled in the art that any disjunctive words and / or phrases representing two or more alternative terms, whether in the specification, claims or drawings, should generally be understood to cover the possibility of including one of the terms, any one of the terms, or two terms, unless otherwise specified in the context. Therefore, the detailed description should not be understood in a restrictive sense, and the scope is limited by the attached claims and their equivalents. In this document, as common in patent documents, the term "one" or "an" is used to include one or more than one. In this document, the term "or" is used to refer to a non-exclusive "or", so that unless otherwise indicated, "A or B" includes "A but not B", "B but not A" and "A and B".

[0102] All patents, patent applications, publications, or other public materials mentioned herein are incorporated herein by reference in their entirety, as if each individual reference was expressly incorporated by reference. All references and any materials or portions thereof referred to as incorporated herein by reference are incorporated herein only to the extent that the incorporated materials do not conflict with existing definitions, statements, or other public materials set forth in the present disclosure. Thus, and to the extent necessary, the disclosure set forth herein supersedes any conflicting materials incorporated herein by reference, and the disclosure expressly set forth in the present application shall prevail.

[0103] Those skilled in the art will recognize that, in general, the terms used herein and particularly in the appended claims (e.g., the appended claim bodies) are generally intended as "open" terms (e.g., the term "including" should be interpreted as "including but not limited to", the term "having" should be interpreted as "having at least", the term "includes" should be interpreted as "includes but is not limited to", etc.). Those skilled in the art will further understand that if a specific number of an introduced claim recitation is intended, such intent will be expressly stated in the claim, and in the absence of such a statement, no such intent is present. For example, to aid understanding, the appended claims may contain the introductory phrases "at least one" and "one or more" used to introduce claim recitations. However, the use of such phrases should not be construed as implying that the introduction of a claim recitation by the indefinite article "a" or "an" limits any particular claim containing such introduced claim recitation to claims containing only one such recitation, even when the same claim includes the introductory phrases "one or more" or "at least one" and indefinite articles such as "a" or "an" (e.g., "a" and / or "an" should generally be construed to mean "at least one" or "one or more"); the same is true for the use of definite articles used to introduce claim recitations.

[0104] Additionally, even if a specific number of an introduced claim recitation is explicitly stated, those skilled in the art will recognize that such recitation should generally be interpreted to mean at least the stated number (e.g., simply stating "two recitations" without other modifiers generally means at least two recitations, or two or more recitations). Furthermore, in those cases where a convention similar to "at least one of A, B, and C, etc." is used, generally such construction is intended in the sense that those skilled in the art would understand the convention (e.g., "a system having at least one of A, B, and C" would include, but is not limited to, systems having only A, only B, only C, A and B together, A and C together, B and C together, and / or A, B, and C together, etc.). In those cases where a convention similar to "at least one of A, B, or C, etc." is used, generally such construction is intended in the sense that those skilled in the art would understand the convention (e.g., "a system having at least one of A, B, or C" would include, but is not limited to, systems having only A, only B, only C, A and B together, A and C together, B and C together, and / or A, B, and C together, etc.).

[0105] With respect to the appended claims, those skilled in the art will appreciate that the operations described therein may be performed in any order in general. In addition, although the claims are presented in one or more sequences, it is understood that the various operations may be performed in an order other than the order described, or may be performed simultaneously. Examples of such alternative orderings may include overlapping, interleaved, interrupted, reordered, incremental, preparatory, supplementary, simultaneous, inverted, or other variations of ordering, unless the context dictates otherwise. In addition, terms such as "responsive to," "related to," or other past tense adjectives are generally not intended to exclude such variations, unless the context dictates otherwise.

[0106] It is worth noting that any reference to "one aspect," "an aspect," "an example," "an example," etc. means that a particular feature, structure, or characteristic described in conjunction with that aspect is included in at least one aspect. Thus, the phrases "in one aspect," "in an aspect," "in an example," and "in an example" appearing in various places throughout this specification are not necessarily all referring to the same aspect. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more aspects.

[0107] As used herein, the singular forms "a," "an," and "the" include plural referents unless the context clearly dictates otherwise.

[0108] Directional phrases used herein, such as, but not limited to, top, bottom, left, right, lower, upper, front, rear and variations thereof, shall refer to the orientation of elements shown in the drawings and shall not be limiting on the claims unless expressly stated otherwise.

[0109] Unless otherwise specified, the term "about" or "approximately" used in this disclosure means an acceptable error for a particular value determined by one of ordinary skill in the art, which depends in part on how the value is measured or determined. In some aspects, the term "about" or "approximately" means within 1, 2, 3, or 4 standard deviations. In some aspects, the term "about" or "approximately" means within 50%, 200%, 105%, 100%, 9%, 8%, 7%, 6%, 5%, 4%, 3%, 2%, 1%, 0.5%, or 0.05% of a given value or range.

[0110] In this specification, unless otherwise indicated, all numerical parameters should be understood as being preceded and modified in all instances by the term "about", wherein the numerical parameters are characterized by the inherent variability of the underlying measurement technology employed to determine the numerical value of the parameter. At the very least, and not as an attempt to limit the application of the doctrine of equivalents to the scope of the claims, each numerical parameter described herein should at least be construed in light of the number of reported significant digits and by applying ordinary rounding techniques.

[0111] Any numerical range listed herein includes all subranges that fall within the listed range. For example, the range "1 to 100" includes all subranges between the listed minimum value 1 and the listed maximum value 100 (and including the minimum and maximum values), that is, having a minimum value equal to or greater than 1 and a maximum value equal to or less than 100. In addition, all ranges listed herein include the endpoints of the listed ranges. For example, the range "1 to 100" includes endpoints 1 and 100. Any maximum numerical limit stated in this specification is intended to include all lower numerical limits contained therein, and any minimum numerical limit stated in this specification is intended to include all higher numerical limits contained therein. Therefore, the applicant reserves the right to modify this specification (including claims) to explicitly state any subranges contained within the range explicitly stated herein. All such ranges are inherently described in this specification.

[0112] The terms "comprise" (and any form of including, such as "comprises" and "comprising"), "have" (and any form of having, such as "has" and "having"), "include" (and any form of including, such as "includes" and "including"), and "contain" (and any form of containing, such as "contains" and "containing") are all open linking verbs. Thus, a system that "comprises," "has," "includes," or "contains" one or more elements possesses those one or more elements, but is not limited to possessing only those one or more elements. Similarly, an element of a system, apparatus, or device that "comprises," "has," "includes," or "contains" one or more features possesses those one or more features, but is not limited to possessing only those one or more features.

[0113] The corresponding structures, materials, actions and equivalents of all members or step-plus-function elements in the following claims are intended to include any structure, material or action for performing a function in combination with other claimed elements as specifically claimed. The description of the technology of the present invention has been presented for the purpose of illustration and description, but this description is not intended to be exhaustive and is not limited to the disclosed form. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope of the claimed subject matter. Exemplary aspects are selected and described to best explain the principles of the technology of the present invention and its practical application, and to enable those of ordinary skill in the art to understand the various aspects of the present disclosure with various modifications suitable for the intended specific use.

Claims

1. An automated computer-implemented method of determining information sentiment in digital information, the method comprising: deriving, by at least one processor, digital information from a source; generating, by the at least one processor, a domain-specific machine-learned sentiment score based on the digital information using one of at least two machine-learned models; autonomously drawing, by the at least one processor, a non-domain-specific knowledge graph of associations between elements in a set of digital context information; receiving, by the at least one processor, an emotion map, each emotion map in the emotion map defining an emotion; generating, by the at least one processor, a graph sentiment score based on the non-domain-specific knowledge graph and the sentiment graph; generating, by the at least one processor, a final sentiment score based on the atlas sentiment score and the domain-specific machine learning sentiment score; as well as The information sentiment in the digital information is determined by the at least one processor based on the final sentiment score.

2. The method according to claim 1, further comprising: Entity attributes in at least one of a database or a server are automatically updated by the at least one processor based on the final sentiment score.

3. The method according to claim 1, further comprising: training, by the at least one processor, a first machine learning model having a base layer and a second layer on the digital information; incorporating, by the at least one processor, the base layer trained on the digital information into a second machine learning model; as well as The second machine learning model including the base layer and the final layer is trained by the at least one processor to generate the domain-specific machine learning sentiment score.

4. The method of claim 3, wherein the training of the first machine learning model comprises training the first machine learning model to classify the subject matter of the digital information.

5. The method of claim 1, wherein the generating of the graph sentiment score comprises: Determining, by the at least one processor, a graph similarity between each emotion graph in the emotion graph and the non-domain-specific knowledge graph; applying, by the at least one processor, the emotion defined by each of the emotion graphs to its determined graph similarity to generate a graph-specific similarity tone score; as well as The graph-specific similarity tone scores for each of the emotion graphs are combined, by the at least one processor.

6. The method of claim 1 , wherein the generating of the final sentiment score comprises: applying, by the at least one processor, a weighting to the graph sentiment score to generate a weighted graph sentiment score; applying, by the at least one processor, another weighting to the domain-specific machine-learned sentiment score to generate a weighted domain-specific machine-learned sentiment score; as well as The weighted graph sentiment score and the weighted domain-specific machine learning sentiment score are combined, by the at least one processor. The method of claim 1 , wherein the element comprises at least one of an entity, a name, a location, a time, or an event. The method of claim 1 , wherein at least a portion of the digital information is labeled.

9. The method of claim 1, wherein the emotions defined by each of the emotion graphs are related to a digitally provided background context.

10. An automated system for updating stored entity attributes based on determined information sentiment, the automated system comprising: A database, which contains entity attributes; at least one processor; as well as A computer readable medium storing instructions executable by the processor to: inputting, by the at least one processor, domain-specific digital information received from a source into a trained domain-specific machine learning model; outputting, by the at least one processor, a domain-specific sentiment score generated by the trained domain-specific machine learning model; Inputting the digital news information into a knowledge graph representing an entity by the at least one processor; Updating the knowledge graph with the digital news information by the at least one processor; Determining, by the at least one processor, a similarity between the knowledge graph and a defined sentiment graph to generate a graph sentiment score; generating, by the at least one processor, an entity sentiment score based on the domain-specific sentiment score and the graph sentiment score; searching, by the at least one processor, an entity sentiment score entry stored in the database; and The entity sentiment score entry in the database is automatically updated by the processor based on a difference between the entity sentiment score and the entity sentiment score entry.

11. An automated system according to claim 10, wherein the automatic update of the entity sentiment score entry in the database includes at least one of: deleting, changing, adding to, subtracting from, or applying a weight to the entity sentiment score entry in the database.

12. A connection system consisting of a cluster of nodes to create and update entity profiles based on real-time information, the connection system comprising: a plurality of nodes connected within the cluster; A first node of the plurality of nodes in communication with the digital information channel, the first node comprising instructions executable to perform the following operations: receiving digital information associated with an entity from the digital information channel; inputting the digital information into a trained machine learning (ML) network to generate a sentiment classification; Outputting the emotion classification to a processing node among the plurality of nodes; as well as A second node among the plurality of nodes in communication with a news source, the second node comprising instructions to perform the following operations: receiving digital news content from said news source; as well as A knowledge graph associated with the entity is drawn based on the digital news content.

13. The connection system of claim 12, wherein the second node is further in communication with at least one domain user server to receive a sentiment classification of content from the domain user server, wherein the sentiment classification is generated by a domain user.

14. The connection system of claim 13, wherein the domain user server includes instructions to: receiving a new emotion classification from at least one domain user; and The user emotion database storing the emotion classification is updated with the new emotion classification received from the at least one domain user.

15. The connection system of claim 13, wherein the second node includes further instructions to: receiving the emotion classification from the domain user server; and A mood map is drawn based on the mood classification, wherein each mood map contains a mood tone.

16. The connection system of claim 15, wherein the second node includes further instructions to: generating an entity emotion score based on a similarity between the knowledge graph and at least one of the emotion graphs; and The entity sentiment score is output to the processing node.

17. The connection system of claim 16, wherein the processing node includes instructions to: receiving the emotion classification from the first node; receiving the entity sentiment score from the second node; determining a final entity emotion score based on the emotion classification and the entity emotion score; as well as The final entity sentiment score is pushed to a database node among the plurality of nodes, the database node storing a profile of the entity.

18. The connection system of claim 17, wherein the database node includes instructions to: receiving the final entity sentiment score from the processing node; and The profile of the entity stored in the database node is updated with the final entity sentiment score.

19. The connection system of claim 12, wherein the first node includes instructions to: The digital information is detected from the digital information channel.

20. The connection system of claim 12, wherein the second node includes instructions to: The digital news content is detected from the news source.

Images