# US 2020/0401932 A1 · Automated enhancement of opportunity insights > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2020/0401932 A1 - Document ID: US20200401932A1 - Status: Published application - Kind: A1 - Application number: 16/447,849 - Filed: 2019-06-20 - Published: 2020-12-24 - Earliest claimed: 2019-06-20 - Assignees: Oracle International Corporation - Applicants: Oracle International Corporation - Inventors: Vivek Kumar; Catherine You Francis; Meeten Bhavsar - Web page: https://patentgrep.com/patent/US20200401932A1 ## CPC - G06Q 30/0201: Market modelling; Market analysis; Collecting market data - G06F 16/953: Querying, e.g. by the use of web search engines - G06N 3/09: Supervised learning - G06N 20/00: Machine learning - G06Q 10/40: Business processes related to social networking or social networking services - G06Q 30/0202: Market predictions or forecasting for commercial activities - G06Q 30/0631: {Recommending goods or services} - G06N 3/08: Learning methods - G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks - G06N 20/10: using kernel methods, e.g. support vector machines [SVM] - G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract Techniques for enhancing actionable opportunities through machine learning are disclosed. In some embodiments, a system includes an event listener for detecting an opportunity to perform an action using one or more computing resources. When a new opportunity is detected, the system generates a set of search criteria, which is used to search a set of external web resources for current events. The system may then generate, using a machine learning model as a function of one or more features extracted from current events satisfying the search criteria, a score representing a likelihood of success that the action leads to an optimal result. The system may tune the machine learning model based on feedback received that is indicative of how the current events affected the likelihood of success for the action. ## Claims ### Claim 1 One or more non-transitory computer-readable media storing instructions which, when executed by one or more hardware processors cause: detecting an opportunity to perform an action using at least one computing resource; responsive to detecting the opportunity to perform the action using the at least one computing resource, generating a set of search criteria to search for current events related to the opportunity; searching a set of external web resources for current events satisfying the set of search criteria; generating, using a machine learning model based on at least one current event that satisfies the set of search criteria, a score representing a likelihood of success for the action using the at least one computing resource; receiving feedback that is indicative of how the at least one current event affects the likelihood of success for the action using the at least one computing resource; tuning the machine learning model for future recommendations based at least in part on the feedback. ### Claim 2 (depends on 1) The one or more non-transitory computer-readable media of claim 1, wherein the instructions further cause: generating sentiment information for the at least one current event; forming a feature vector based at least in part on the sentiment information; wherein generating, using the machine learning model based on at least one current event that satisfies the set of search criteria, the score comprises determining, as a function of the feature vector and the machine learning model, a probability that the action using the at least one computing resource will cause a particular result. ### Claim 3 (depends on 2) The one or more non-transitory computer-readable media of claim 2, wherein the sentiment information is generated using a trained classification model that classifies the at least one current event as having a positive sentiment, a negative sentiment, or a neutral sentiment. ### Claim 4 (depends on 2) The one or more non-transitory computer-readable media of claim 2, wherein the feature vector includes values for one or more features extracted from the sentiment information and one or more features extracted from non-sentiment information about the at least one current event. ### Claim 5 (depends on 1) The one or more non-transitory computer-readable media of claim 1, wherein the instructions further cause: generating a set of tags for the at least one current event; forming a feature vector based at least in part on the set of tags; wherein generating, using the machine learning model based on at least one current event that satisfies the set of search criteria, the score comprises determining, as a function of the feature vector and the machine learning model, a probability that the action using the at least one computing resource will cause a particular result. ### Claim 6 (depends on 1) The one or media of claim 1, wherein the score is further generated, based at least in part, on how much time has lapsed since the opportunity was detected. ### Claim 7 (depends on 1) The one or media of claim 1, wherein the feedback is received by a user through a user interface and indicates whether the at least one current event is relevant to the action using the at least one computing resource. ### Claim 8 (depends on 1) The one or more non-transitory computer-readable media of claim 1, wherein the instructions further cause: receiving, through a user interface, a request to perform the action using the at least one computing resource; responsive to the request, performing the action using the at least one computing resource; wherein the feedback identifies an outcome of the action using the at least one computing resource. ### Claim 9 (depends on 1) The one or more non-transitory computer-readable media of claim 1, wherein the action using the at least one computing resource comprises posting a message on a social media platform. ### Claim 10 (depends on 1) The one or more non-transitory computer-readable media of claim 1, wherein the action using the at least one computing resource comprises sending an electronic message to one or more recipients using one or more web services. ### Claim 11 (depends on 1) The one or more non-transitory computer-readable media of claim 1, wherein the action using the at least one computing resource comprises configuring at least one software or hardware resource. ### Claim 12 (depends on 1) The one or more non-transitory computer-readable media of claim 1, wherein tuning the machine learning model for future recommendations comprises updating at least one weight associated with at least one node or cluster in the machine learning model. ### Claim 13 (depends on 1) The one or more non-transitory computer-readable media of claim 1, wherein the instructions further cause: generating a label for the action based on the feedback; wherein tuning the machine learning model for future recommendations comprises retraining the machine learning model based on the label for the action. ### Claim 14 (depends on 1) The one or more non-transitory computer-readable media of claim 1, wherein the instructions further cause: generating a recommendation for the action based at least in part on the score; presenting the recommendation for the action through a user interface. ### Claim 15 (depends on 1) The one or more non-transitory computer-readable media of claim 1, wherein the instructions further cause: determining, based at least in part on whether the score satisfies a threshold, whether to include the action in a set of recommended actions using the at least one computing resource; presenting the set of recommended actions using the at least one computing resource through a user interface. ### Claim 16 (depends on 1) The one or more non-transitory computer-readable media of claim 1, wherein the instructions further cause: generating summary information for the at least one current event; and presenting, through a user interface, the summary information with a recommendation for the action. ### Claim 17 (depends on 1) The one or more non-transitory computer-readable media of claim 1, wherein the set of one or more external web resources include one or more websites that track current events. ### Claim 18 (depends on 1) The one or more non-transitory computer-readable media of claim 1, wherein the opportunity is associated with a particular entity and the instructions further cause: responsive to detecting the opportunity, determining whether a watchlist including the set of search criteria has already been generated for the opportunity; responsive to determining that the watchlist has not been generated, generating the watchlist for the opportunity, wherein the watchlist periodically or continuously triggers a web crawler to search the set of external web resources for the current events satisfying the set of search criteria; wherein the web crawler searches for current events that affect the particular entity; determining, using a trained classification model, whether the at least one current event that satisfies the set of search criteria has a positive, negative, or neutral impact on the particular entity; forming a feature vector that includes one or more features indicative of whether the at least one current event has a positive, negative, or neutral impact on the particular entity; wherein generating, using the machine learning model based on at least one current event that satisfies the set of search criteria, the score comprises determining, as a function of the feature vector and the machine learning model, a probability that the action using the at least one computing resource will cause a particular result. ### Claim 19 A system comprising: one or more hardware processors; one or more non-transitory computer-readable media storing instructions which, when executed by the one or more hardware processors cause operations comprising: detecting an opportunity to perform an action using at least one computing resource; responsive to detecting the opportunity to perform the action using the at least one computing resource, generating a set of search criteria to search for current events related to the opportunity; searching a set of external web resources for current events satisfying the set of search criteria; generating, using a machine learning model based on at least one current event that satisfies the set of search criteria, a score representing a likelihood of success for the action using the at least one computing resource; receiving feedback that is indicative of how the at least one current event affects the likelihood of success for the action using the at least one computing resource; and tuning the machine learning model for future recommendations based at least in part on the feedback. ### Claim 20 A method comprising: detecting an opportunity to perform an action using at least one computing resource; responsive to detecting the opportunity to perform the action using the at least one computing resource, generating a set of search criteria to search for current events related to the opportunity; searching a set of external web resources for current events satisfying the set of search criteria; generating, using a machine learning model based on at least one current event that satisfies the set of search criteria, a score representing a likelihood of success for the action using the at least one computing resource; receiving feedback that is indicative of how the at least one current event affects the likelihood of success for the action using the at least one computing resource; and tuning the machine learning model for future recommendations based at least in part on the feedback.