Abstract: A method and a system are disclosed for solving a mixed-integer programming problem, the method comprising obtaining an indication of a mixed-integer programming optimization problem; until a performance criterion is met: providing the mixed-integer programming optimization problem to an optimization oracle adapted for solving the mixed-integer programming optimization problem using a feasibility pump technique and comprising an optimization solver, initializing parameters of an optimization oracle and an initial solution pair, the parameters comprising Monte-Carlo simulation parameters, a list of neighborhood functions and a measure of fractionality, and performing iterative calls to the optimization solver until a stopping condition is met; and providing at least one corresponding solution obtained from the optimization solver.

Abstract: Various embodiments are described for dynamic value stream management. A computing environment is directed to receive a stream of metrics from station computing devices each positioned at a station in a manufacturing process, where individual ones of the station computing devices have a sensor configured to generate metrics. The computing environment may determine an optimal allocation of resources for each of the stations in the manufacturing process based at least in part on the metrics. If a cycle time of a station falls below a threshold, personnel from another satisfactorily-performing station may be reassigned to the station based on cross-training metrics. A recommended action for the stations may be determined and presented in a display device.

Abstract: A model requester node, which is an edge node of a cloud computing network, generates a specification of a machine learning model, distributes the specification to a plurality of other edge nodes, and receives replies to the specification from the plurality of other edge nodes. In response to the replies, the model requester node identifies a set of participating edge nodes based on a learning utility and a cost estimate of each of the plurality of other edge nodes. The model requester node then trains the machine learning model, without exchanging training data among the model requester node and the participating edge nodes, by repeatedly: distributing most recent parameters of the machine learning model to the participating edge nodes; receiving updates to the most recent parameters from the participating edge nodes; and establishing new parameters for the machine learning model by aggregating the updates from the participating edge nodes.

Abstract: A method for issuing control signals in response to sentiment. In some embodiments, the method includes: assigning, to each of a plurality of comments, a respective topic vector of length k; determining whether a largest element of the topic vector of a first comment of the plurality of comments exceeds a weight threshold; in response to the determining that the largest element exceeds the weight threshold, classifying the first comment into a first topic, of k topics, the first topic corresponding to the position, in the topic vector, of the largest element of the topic vector; calculating a first sentiment score; calculating an average sentiment score, based in part on the first sentiment score; determining whether the average sentiment score meets a criterion; and in response to the determining that the criterion is met, generating a control signal, the control signal including a message related to the first topic.

Abstract: In an example, a computer-implemented method to structure an analytical workflow that improves reasoning based on a problem context and demonstrated abilities of each individual user may include displaying a reasoning problem to an analyst. The method may include receiving input from the analyst to identify a reasoning problem type of the reasoning problem. providing a recommended analytic approach for the reasoning problem type to the analyst. The method may include assisting the analyst in analyzing and evaluating one or more information sources relevant to the reasoning problem. The method may include guiding the analyst through a structured technique (ST) to support reasoning of the analyst in formulation of a solution to the reasoning problem. The method may include generating a report that includes the analyst's solution to the reasoning problem based on input from the analyst.

Abstract: Disclosed are systems, computer servers, and methods to provide features of an automotive commerce exchange platform to users via a single sign-on. A computer server is configured to provide a plurality of different third-party features of a plurality of different third-party feature providers on the automotive commerce exchange platform, and to provide to the users, via user devices, access to individualized subsets of the third-party features via a plurality of single individualized sets of sign-on credentials. Each single individualized set of sign-on credentials corresponds to a different user.

Abstract: Approaches relating to customer service management for recurring revenue assets are described. A CSM application or other CSM software tool (or set of tools) can include features relating to customer lifecycles, product usage, and/or customer success, each of which can be monitored based on various analytical measures, metrics, indicators, etc. provided by one or more analytical engines. Systems, methods, computer program products, and the like are described.

Abstract: The present application includes a method and system for gathering customer information through games. The system transmits offers to play games over the contact medium used by the customer. The games are selected to elicit information from the customer; information ranging from the customer's mood to marketing information to security information. The information so obtained can be used to update client profiles.

Abstract: There is disclosed a method and system for determining a productivity rate of a user in a computer-implemented crowd-sourced environment. The method comprises, in a training phase: acquiring a training project; determining, set of project-specific features indicative of one or more characteristics of the training project; acquiring, a plurality of training results; determining a set of user-task specific features; determining, a user activity history associated with the user; generating a set of training data including the set of project-specific features, the set of user-task specific features, and the user activity history; training a machine learning algorithm (MLA), the training including: determining, a set of features representative of a property of the set of training data; and generating an inferred function based on the set of features, the inferred function being configured to determine the productivity rate of the user for a given project.

Abstract: Systems and methods for optimizing application experiences on devices in an enterprise while maintaining privacy compliance are provided. An enterprise model management service of the enterprise accesses, from a centralized software service, a model comprising one or more rules to be applied to an application. The model is provided to devices of a first cohort, whereby a rule of the model causes an action associated with the application to occur at each device. User metrics associated with the action at each device of the first cohort is aggregated. The user metrics indicate a result of the action at each device. The aggregated user metrics are analyzed, whereby the analyzing includes determining a second cohort at the enterprise. The model is then provided to devices of at least a portion of the second cohort. No user metrics are returned to the centralized software service thus maintaining privacy compliance.

Abstract: Systems for centralized processing of regulatory control events. A method embodiment applies regulatory compliance rules against regulatory control events that occur at a plurality of heterogeneous remote cloud-based systems. A centralized cloud-based platform manages the compliance of the plurality of heterogeneous remote cloud-based systems by applying a set of data compliance rules pertaining to regulatory controls. The regulatory controls pertain to data access events and data manipulation events that occur on the plurality of computing systems. The centralized cloud-based platform receives control event messages, the control event messages being raised any one or more of the heterogeneous remote cloud-based systems. Rules are processed against the received control event messages to determine a set of compliance actions.

Abstract: A first hierarchy path associated with a first interface of a first provider is determined, with the first interface being of a first type. A second hierarchy path associated with a second interface of the first provider is determined, with the second interface being of a second type different from the first type. The first hierarchy path is determined to be a mismatch from the second hierarchy path. A third hierarchy path corresponding to another interface of the first type is obtained from a second provider. A dictionary is generated based on a determination that the third hierarchy path matches the first hierarchy path. A vector that corresponds to a third interface is generated based on the dictionary. As a result of the third interface being determined to be of the first type based on the vector, a device is caused to perform an operation against the third interface.

Abstract: A method of performing an online assessment of compliance with cyber-security risk and information security risk standards, comprising the steps of providing access to a user to a compliance navigator software tool, the compliance navigator software tool comprising a content management system an assessment module, an access control module, an audit module, and an automated testing module, presenting a plurality of assessment questions to the user by executing the assessment module, assessing answers to the plurality of assessment questions using the assessment module, the assessment module comprising machine-readable instructions stored on the non-transitory machine-readable storage medium, and displaying results of the assessment to the user.

Abstract: Described is a method for generating a score risk (IR), the method comprising the steps of: receiving at least one purchase order for at least one asset (DC), the purchase order of the asset generated from an agreement (A) established between a first entity (C) and a second entity (V), the agreement linked to the performance of an action between the entities. The generation of the agreement (A) also generates an ideal transactional ballast (LI) comprising a plurality of ideal markers wherein each one of the ideal markers is related to the agreement (A) established between the entities (C,V). The performance of the action between the entities generates a real transactional ballast (LR) comprising a plurality of real markers, wherein each one of the real markers is related to the action carried out between the entities, wherein the method further comprises the step of generating and updating the score risk based on the comparison between each one of the real markers and each one of the ideal markers.

Abstract: A modeling server is in communication with multiple databases and/or knowledge systems. The modeling server may utilize predetermined rules for suggesting design elements such as plants, lighting, fencing, irrigation, and more. The modeling server may acquire legal information pertaining to the project, based on jurisdictions that contain the location of the project. The modeling server may further generate one or more teams to implement the project, including, but not limited to, design teams, construction teams, and/or management teams. The modeling server may further produce a rendering and/or simulation of a landscaping or landscape architecture project to facilitate improved customer engagement, which may result in increased customer satisfaction as well as improved project performance in terms of scheduling, budget, and longevity of the project.

Abstract: A method for operating a self-orchestrated system for extraction, analysis, and presentation of entity data involves extracting a web page to object-based storage including web page content, web page metadata and a globally unique identifier. The method extracts the web page metadata from the object-based storage. The method inputs the web page metadata to a queue. The method pulls web page content from a content store. The method receives RegEx from a model parameter store. The method parses the web page content using RegEx and web page metadata. The method passes web page metadata and extracted content from the web page and positions of extracted content to an advanced analysis function decider (AAF Decider) for analysis. The method streams web page metadata and extracted content from the web page and positions of the extracted content to a JSON file batch for flattening.

Abstract: A printer includes: a printing section configured to perform printing using a plurality of printing materials contained in corresponding printing material containers, a display section configured to display respective remaining amounts of the individual printing materials contained in the corresponding printing material containers; and a controller configured to control display on the display section, wherein the controller performs display control so as to match the display on the remaining amount of a second container other than the first container with the display on the remaining amount of a first container having a smallest remaining amount among the plurality of the printing material containers.

Abstract: A system that provides a safety rating for a driver or a vehicle to a rider of the vehicle. Information including diagnostic information such as sensor information and location information is used to determine the safety rating. By having a high safety rating, the rider of the vehicle has some level of assurance that the vehicle will arrive at a designated location.

Abstract: A portable handheld device receives from a central repository, information on a failed hardware component of a computational device, wherein the information includes an authentication code to permit access to the failed hardware component and a time window in which the failed hardware component is permitted to be accessed. The portable handheld device uses the authentication code to access the failed hardware component for repair or replacement during the time window.

Abstract: Information is gathered on microservice interactions. Two or more microservice failures are detected. For each microservices failure, a microservice restoration time is determined. An expected total cost of a downtime for each microservice is determined. Based on the determined expected total cost of the downtime for each microservice, an order of microservices to restore is determined.

Abstract: The present disclosure describes systems and methods for identifying products that are implicitly published in an image, e.g., where there is no identifying text or comments that identify the brand of the product. A product entity may be identified based on comments posted in response to the posting of the image. An analysis of the image may be performed based on the identified product entity to find items or products of a type of the identified product entity in the image. A product detected in the image may be compared to a database of products of the type of product entity to determine which product was posted in the image.

Abstract: A social intelligence system is presented that streams information from a source, queues the streamed information, analyzes/scores the queued data, and stores the analyzed/scored data in an analysis database. The analyzed/scored data can then be retrieved from the database for post-processing and stored in a client specific database for further reporting. By streaming the data into various message queues and scoring the data before storing in the analysis database, large volumes of data can be efficiently processed and analyzed for a particular person and/or entity.

Abstract: Mediums, apparatus, computer program code, and means may be provided to evaluate relative risks based at least in part on source inputs received via a distributed communication network by an automated back-end application computer server. According to some embodiments, the server may access a data store containing electronic files associated with a set of entities to retrieve, for each of a plurality of the entities in the set of entities, electronic files associated with that entity. The server may also retrieve structured data elements, unstructured data elements, and external, third-party data elements for that entity. The server may then execute an automated risk model to assign a risk score to that entity based on the electronic files, the structured data elements, the unstructured data elements, and the external, third-party data elements for that entity and transmit indications of the risk scores for the plurality of entities.

Abstract: A method and system for creating a data grouping platform for grouping data from disparate locations and sources, according to a user-defined aggregate. The method and system includes a method for receiving a user-defined aggregate, a user-defined trigger event, and a plurality of user-defined grouping criteria and parameters for creating the user-defined aggregate. The system and method also allows for a user to identify input data sets and sources for creating the grouped aggregates. Once the method and system receives the user-defined parameters, the method and system builds a platform to group data and build data aggregates based on the user-defined criteria.

Abstract: A retrieving device includes: a classifying unit that classifies document data made up of a plurality of items on the basis of a degree of matching of a first item in which a combination of one or more symbol strings is described in a database in which the document data is stored; and an extracting unit that extracts a group of keywords included in a second item from pieces of document data in a same group classified by the classifying unit as related words for retrieval. Moreover, when the retrieving device is implemented as a distributed system on a network, it is possible to create related words for a plurality of machines and to store the related word data and to enable a plurality of users to use the distributed system.

Abstract: Embodiments discussed herein refer to systems and methods for identifying relevantly similar sentiment in text strings.

Abstract: A method for continuously tracking business performance impact of order sourcing systems and algorithms that decide how ecommerce orders should be fulfilled by assigning the items of the order to nodes in a fulfillment network such as stores, distribution centers, and third party logistics—to provide automatic root cause analysis and solution recommendations to pre-defined business problems arising from KPI monitoring. A Business Intelligence (BI) dashboard architecture operates with: 1) a monitoring module that continuously monitors business KPIs and creates abnormality alerts; and 2) a root cause analysis module that is designed specifically for each business problem to give real time diagnosis and solution recommendation. The root cause analysis module receives the created alert, and triggers conducting a root cause analysis at an analytics engine. The BI dashboard and user interface enables visualization of the KPI performance and root cause analysis results.

Abstract: A system and method provide the ability to control an architecture, engineering, and construction (AEC) project workflow. AEC data regarding a quality of construction is obtained. A set of classifiers and machine learning models are obtained. The AEC data is augmented based on the set of classifiers and machine learning models. A risk metric is generated for one or more issues in the AEC data based on the augmented AEC data. The risk metric is interactively generated and presented on a display device. Work, project resourcing, and/or training are prioritized based on the risk metric.

Abstract: A system and method for identifying multiple excavation sites and generating multiple locate requests from a single locate request session comprising an electronic interactive interface for defining a geographic area, wherein the geographic area defined represents an area for evaluation of the presence or absence of one or more underground facilities, an auto-populating interactive smart map, a rules engine for determining a number of tickets per area defined, and a database and/or electronic storage of one or more underground facilities.

Abstract: An image display device includes an approach situation information acquirer that acquires approach situation information indicating approach of at least part of a customer's body to a shelf on which a plurality of items are displayed and indicating, when there is the approach, a position of the approach on the shelf at each time; an associator that associates the approach situation information with one item of the plurality of items according to the position indicated by the approach situation information; an index value calculator that calculates an index value relating to the customer's behavior on the basis of information indicating the item associated with the approach situation information; and a display unit that displays the index value calculated by the index value calculator together with an image of the item on the basis of the association made by the associator.

Abstract: Crowdsourcing systems with machine learning are described. Specifically, item-label inference methods and systems are presented, for example, to provide aggregated answers to a crowdsourced task in a manner achieving good accuracy even where observed data about past behavior of crowd members is sparse. In various examples, an item-label inference system infers variables describing characteristics of both individual crowd workers and communities of the workers. In various examples, an item-label inference system provides aggregated labels while considering the inferred worker characteristics and the inferred characteristics of the worker communities. In examples the item-label inference system provides uncertainty information associated with the inference results for selecting workers and generating future tasks.

Abstract: A system, method and computer program product for continuously tracking business performance impact of order sourcing systems and algorithms that decide how ecommerce orders should be fulfilled by assigning the items of the order to nodes in a fulfillment network such as stores, distribution centers, and third party logistics—to provide automatic root cause analysis and solution recommendations to pre-defined business problems arising from KPI monitoring. A Business Intelligence (BI) dashboard architecture operates with: 1) a monitoring module that continuously monitors business KPIs and creates abnormality alerts; and 2) a root cause analysis module that is designed specifically for each business problem to give real time diagnosis and solution recommendation. The root cause analysis module receives the created alert, and triggers conducting a root cause analysis at an analytics engine. The BI dashboard and user interface enables visualization of the KPI performance and root cause analysis results.

Abstract: A method and apparatus for managing a networked vehicle resource sharing facility, the method and system provide for: defining a geographical area as a work region; storing, in a data structure, data relating to vehicle hire bookings beginning inside the work region, wherein the booking information for each booking includes a default delay time; applying a rule to determine that an additional delay condition or fully booked condition is satisfied within the work region; and in response to positively determining, automatically introducing an additional delay or a fully booked setting, respectively, to at least some of the vehicle hire bookings inside the work region.

Abstract: Aspects described herein pertain to the use of an adaptive and tunable processing system and method to provide a risk assessment and compliance analysis, but are not so limited. An adaptive and tunable processing system of an embodiment adaptively determines an amount of compliance and/or risk based on one or more quantification variables, including one or more of a litigation category variable, a licensing category variable, and/or a consumer sentiment category variable, to generate a stratified score that quantifies an amount of compliance and/or risk associated with a particular entity. The system of an embodiment uses one or more adaptive filters tuned to output values corresponding to an amount of compliance and/or risk associated with use of a particular entity to perform a service or services.

Abstract: In some embodiments, apparatuses and methods are provided herein useful for a retail shopping facility to maintain stocked shelves, in part, from obtaining images of the store shelves via cameras mounted onto shopping carts. By one approach, the images obtained from the shopping carts are compared with planogram images to determine where retail products are needed to restock the shelves. In some examples, the shopping cart has a control circuit that determines when to capture electronic images based, in part, upon a sensor mounted onto the shopping cart, instructs the camera to capture the electric image, and stores the captured electronic image in the memory.

Abstract: Systems, apparatuses, and methods for modular compliance monitoring are provided. A system for monitoring modular compliance in a shopping space includes an image capture device, a baseline condition database, and a control circuit coupled to the image capture device and the baseline condition database. The control circuit being configured to obtain, from the image capture device, one or more images of a display module in the shopping space, retrieve, from the baseline condition database, a baseline condition model for a display space corresponding to the display module, and compare the one or more images of the display module in the shopping space and the baseline condition model to determine a modular compliance status for the display module.

Abstract: In one embodiment, a method includes accessing content objects of an online social network, each content object being associated with an entity of the online social network, where each content object includes content of the content object and is associated with metadata, generating a set of n-grams by extracting one or more n-grams from the content of the content object, calculating, for each extracted n-gram, a quality score for the n-gram based on occurrence counts associated with map tiles of a geographical map, where each occurrence count comprises a count of entities geographically located in a corresponding map tile and associated with the n-gram, generating a snippet-module including one or more of the extracted n-grams from the set of n-grams having quality-scores greater than a threshold quality-score, and sending, to a client system of a user of the online social network, the snippet-module for display to the user.

Abstract: The disclosure includes a system and method for reconfiguring a vehicle based on one or more preferences of a first user without the first user directly providing an input to the vehicle to reconfigure the vehicle. The method may include wirelessly receiving first user profile data at a vehicle. The first user profile data may be associated with a first user. The first user profile data may describe how one or more settings of the vehicle should be configured for the first user. Upon receipt of the first user profile data, the vehicle may be configured in accordance with second user profile data describing how the one or more settings of the vehicle should be configured for a second user. The method may include reconfiguring the one or more settings of the vehicle based on the first user profile data so that the vehicle is reconfigured for the first user.

Abstract: A device, method and computer-readable medium for providing failure management of a data streaming processing system are disclosed. A method may detect a first replica of a shard has failed during processing of the stream of data records between a first operator and a second operator, notify the first operator and the second operator that the first replica of the shard has failed, perform a rewind procedure to identify data records that have been processed in the first replica and resume processing of the stream of data records between the first operator and the second operator via a second replica of the shard without processing the data records that have already been processed from the first replica.

Abstract: A method of tracking and communicating information about a person to a provider of goods or services is disclosed. A GPS enabled mobile phone is used to search for nearby sites of providers of a specific service. Once a service provider is selected, a route to that provider is advised and as the person travels towards the provider the separation between them is determined. As one or more geofences are broken during the journey warning messages are sent to the service provider indicating the pending arrival. The method is particularly useful where the service is to be provided to a disabled person as their accessibility needs can be identified in the message.

Abstract: The on-boarding framework consolidates and ensures consistent implementation of business compliance with regulatory policy across a global business that is required to comply with regulations of international, regional, country specific, local jurisdictions, or the like. The on-boarding framework facilitates the work flow, approvals, documents, reporting, and other information using dynamic rules that dictate the on-boarding elements based on customer type, booking entity, location of the authorized approval, special products, data collected, and the like, which are used to dynamically evaluate the client's on-boarding potential regulatory issues, and the actions used to mitigate potential regulatory issues with the entity being on-boarded.

Abstract: Mediums, apparatus, computer program code, and means may be provided to evaluate relative risks based at least in part on source inputs received via a distributed communication network by an automated back-end application computer server. According to some embodiments, the server may access a data store containing electronic files associated with a set of entities to retrieve, for each of a plurality of the entities in the set of entities, electronic files associated with that entity. The server may also retrieve structured data elements, unstructured data elements, and external, third-party data elements for that entity. The server may then execute an automated risk model to assign a risk score to that entity based on the electronic files, the structured data elements, the unstructured data elements, and the external, third-party data elements for that entity and transmit indications of the risk scores for the plurality of entities.

Abstract: An end-to-end “pilot as service” system for software pilot testing, comprising a server including at least one processor and operative for generating a set of software testing environment/s simultaneously serving plural clients from among a first plurality of enterprise clients, the environment being operative for concurrently running plural proof-of-concept tests for respective plural software products each being tested (“pilots”), on behalf of an individual start-up client from among a second plurality of start-up clients, by an individual one of the population of enterprise clients; a registration functionality, on the processor, operative for pre-registering at least one of: enterprise clients from among the first plurality of enterprise clients, and start-up clients from among the second plurality of start-up clients; and a dashboard functionality, on the processor, operative for presenting at least one result of at least one proof-of-concept test to each client which has pre-registered using said regist

Abstract: A system and a method for mobile sensing data processing are provided. The method includes, receiving one or more requests from one or more applications installed at a client device to obtain a processed sensing data obtained in response to execution of one or more tasks by the application using a set of sensors. Raw data is extracted from the set of sensors in response to the execution of the tasks. A data stream is configured to include sensor data and a task information associated with the tasks. The client device is connected with the server to transmit the data stream. The server outputs the processed sensing data upon processing the data stream and the task information by using one or more task specific models stored at the server. The processed sensing data is received from the server and provided to the applications.

Abstract: A contextual analysis engine systematically extracts, analyzes and organizes digital content stored in an electronic file such as a webpage. Content can be extracted using a text extraction module which is capable of separating the content which is to be analyzed from less meaningful content such as format specifications and programming scripts. The resulting unstructured corpus of plain text can then be passed to a text analytics module capable of generating a structured categorization of topics included within the content. This structured categorization can be organized based on a content topic ontology which may have been previously defined or which may be developed in real-time. The systems disclosed herein optionally include an input/output interface capable of managing workflows of the text extraction module and the text analytics module, administering a cache of previously generated results, and interfacing with other applications that leverage the disclosed contextual analysis services.

Abstract: In a method of loading checked luggage on unit load devices corresponding to an aircraft having a passenger compartment, each piece of checked luggage is associated with a unique identifier. Each piece of checked luggage is weighed and its weight is associated with its corresponding unique identifier. A best possible luggage placement is calculated considering a distribution of passenger weight, thereby allowing the passengers to sit at their preferred location rather than by pre-allocation. A desired placement of the checked luggage is calculated based on the weight of each piece of checked luggage, the acceptable distribution of weight in the passenger compartment and overall fuel state over mission profile.

Abstract: A contextual analysis engine systematically extracts, analyzes and organizes digital content stored in an electronic file such as a webpage. Content can be extracted using a text extraction module which is capable of separating the content which is to be analyzed from less meaningful content such as format specifications and programming scripts. The resulting unstructured corpus of plain text can then be passed to a text analytics module capable of generating a structured categorization of topics included within the content. This structured categorization can be organized based on a content topic ontology which may have been previously defined or which may be developed in real-time. The systems disclosed herein optionally include an input/output interface capable of managing workflows of the text extraction module and the text analytics module, administering a cache of previously generated results, and interfacing with other applications that leverage the disclosed contextual analysis services.

Abstract: A computer implemented method, implemented by an enterprise, of receiving regulatory change data from a feed, identifying a regulatory change, and communicating the regulatory change to an affected regulatory inventory associated with the enterprise. The method may also comprise of providing a questionnaire to a user, receiving questionnaire answers from a user, and creating a compliance action plan to assist the enterprise in managing the regulatory change.

Abstract: A security platform employs a variety techniques and mechanisms to detect security related anomalies and threats in a computer network environment. The security platform is “big data” driven and employs machine learning to perform security analytics. The security platform performs user/entity behavioral analytics (UEBA) to detect the security related anomalies and threats, regardless of whether such anomalies/threats were previously known. The security platform can include both real-time and batch paths/modes for detecting anomalies and threats. By visually presenting analytical results scored with risk ratings and supporting evidence, the security platform enables network security administrators to respond to a detected anomaly or threat, and to take action promptly.

Abstract: A system and method for analysis of interactions with a customer service center, comprising receiving a plurality of customer service interactions, receiving a word cloud computed for the plurality of customer service interactions, consolidating similar or synonymous words into word sets, constructing a weighted graph of the word sets, generating a query for the interaction topic based on a corresponding subset of word sets, selecting at least one representative interaction from the retrieved customer service interactions, and displaying at least a portion of the representative interaction for the selected identified topic.