Abstract: Systems and methods for on demand local commerce are described. One example embodiment includes a device gathering location information and product interest associated with clients and client devices. The system may use location information in determining that the first plurality of client devices are within a first geographic area during a first time period, and may further use the interest information in calculating an interest level for a first product. A threshold may be identified and used in determining that the interest level for the first product exceeds the threshold. When the calculated interest level exceeds the threshold, a local commerce action is initiated. In various embodiments, the local commerce action may be a live on demand auction at a particular location, an offer associated with a geofenced area, a sales location recommendation to a merchant, or any other such local commerce action.

Abstract: Computerized methods and systems for three-dimensional (3-D) displaying and navigating of search results are provided. In some embodiments, a criterion may be received from a user of a publication system. A database associated with the publication system may be searched based on the criterion. Search results retrieved from the database may be displayed on at least one side of a 3-D interface object. At least one control may be displayed proximate to the three 3-D interface object. The control may be used to navigate the 3-D interface object.

Abstract: An autonomous item generation system implements a trained machine learning model configured to output fabrication instructions for generating an item and metadata describing the item, automatically and independent of user input. Fabrication instructions output by the machine learning model are transmitted to a fabrication device for generating the item. The autonomous item generation system generates a listing for the item based on the metadata output by the machine learning model and publishes the listing to a virtual marketplace. Analytics data describing feedback for the item listing is used to generate training data for the machine learning model. The training data is input to the machine learning model, which causes the machine learning model to refine at least one control parameter according to a loss function that penalizes negative differences between predicted and observed feedback data for the item.

Abstract: A method for compressing data in a local cache of a web server is described. A local cache compression engine accesses values in the local cache and determines a cardinality of the values of the local cache. The local cache compression engine determines a compression rate of a compression algorithm based on the cardinality of the values of the local cache. The compression algorithm is applied to the cache based on the compression rate to generate a compressed local cache.

Abstract: Systems and methods provide determining listings of items based on similarities at least among items and queries in an online shopping system. In particular, the systems and methods determine similarities among items, users, product, messages, reviews, and queries, based on a combination of a machine learning model and similarity index data. The machine learning model (e.g., a Transformer model and a neural network model) generates embedded vector representation of items, queries, and other data in the online shopping systems. The machine learning model may be pre-trained based at least on data associated with items in the online shopping system, and fine-tuned based on a variety of mappings of similarities: item-to-item, user-to-item, query-to-item, and the like. The similarity index data include k-Nearest Neighbor index data for determining items within a range of similarity based on a receive query.

Abstract: Methods for face detection to address privacy in publishing image datasets is described. A method may include face classification in an online marketplace. A server system may receive, from a seller user device, a listing including an image for the online marketplace. The server system may classify, by at least one processor that implement a distribution-balance trained machine learning model, each human face candidate within the image as being one of a private human face or a non-private human face. The server system may receive, from a buyer user device, a search query that is mapped to the listing in the online marketplace. The server system may transmit, to the buyer user device, a query response including the listing that includes the image determined to not include any private human faces or obscures any private human faces within the image based on the classifying.

Abstract: Sequential selections are presented by receiving an input on an active portion of a display of the computing device. While the input is received, a plurality of selections are sequentially displayed on the display. An indication is received that the input on the active portion of the display of the computing device has changed. In response to receiving the indication that the input has changed, the computing device stops presenting the plurality of selections and determines which selection of the plurality of selections was presented on the display when the input changed. The determined selection is then displayed.

Abstract: Methods, systems, and programs are presented for securing user-address information. A first memory is configured according to a first table that does not include information about user identifiers. Each entry in the first table includes a physical location identifier and information about a physical location. A second memory is configured according to a second table, where each entry in the second table includes the physical location identifier and an account identifier of a user for accessing a service. The first and second tables are configured to separate profile information from the address information of the user. Additionally, a firewall is configured to control access to the second memory. The firewall defines an authentication zone including the second memory but not the first memory, where access to the second memory by internal services is allowed and direct access by the user to the second memory is denied.

Abstract: An example embodiment includes a system server configured for social network-based inventory management. The system server includes processors and non-transitory computer-readable storage media communicatively coupled to the processors. The media store instructions that, in response to execution by the processors, cause the processors to perform operations. The operations include receiving category specifications and access category definitions for access categories. The access category definitions include relationship criteria for the access categories. The operations include identifying a relationship between a second user and a first user, comparing an identified relationship with the relationship criteria, and determining which of the access categories the second user is included based on the comparison.

Abstract: Adding additional value to NFTs is described. An association of a non-fungible token (NET) with a user account is verified based on an address of a digital wallet that corresponds to the user account and that is encoded into the NET stored on a blockchain. After verifying the association of the NET with the user account, a condition relative to the NET is detected. Responsive to detecting the condition, ownership of an earned item is conferred to the user account. An NFT collection is formed by digitally bundling the NFT and the earned item, and a listing of the NFT collection is generated for the user account. The listing specifies that the NFT collection includes both the NFT and the earned item.

Abstract: Example embodiments of the present disclosure include a system comprising a computer-readable storage medium storing at least one program and a computer-implemented method for providing an interactive and social fitting room experience. Consistent with some embodiments, the method may include selecting a target audience to poll for feedback on an image of an individual wearing a garment in a fitting room. The method may further include transmitting the image to a computing device of each member of the target audience, and compiling feedback regarding the image from members of the target audience. The method may further include causing the compiled feedback to be displayed to the individual on an interactive mirror display in the fitting room.

Abstract: A physical storage vault for physical items of digital twin NFTs is described. A combined listing for a physical item and an NET of the physical item is generated, and the combined listing is output in a user interface at a client device. The user interface includes an option to store the physical item in a physical storage vault. A request is received from the client device to transfer ownership of the physical item and the NFT of the combined listing to a user account. The request indicates that the option to store the physical item in the physical storage vault is accepted. Responsive to the request, transfer of ownership of the NET to the user account is initiated by encoding metadata of the NFT on a blockchain with an identifier associated with the user account and by storing the physical item in the physical storage vault.

Abstract: An apparatus is provided including a processor, a memory, and a vehicle tracking portal. The vehicle tracking portal is configured to obtain vehicle identification information regarding a vehicle. Then diagnostic information regarding the vehicle can be obtained. Then, based on the vehicle identification and the diagnostic information, a recommended service for the vehicle can be determined. A user can then be alerted to the recommended service.

Abstract: Deploying a point of presence (PoP) changes traffic flow to a cloud service provider. To determine if the PoP improves the performance of a cloud service to a client, actual network latencies between the client and the cloud service are measured. In more complex scenarios, multiple PoPs are used. The client sends multiple requests for the same content to the cloud provider. The requests are sent via different routes. The cloud provider serves the requests and collates the latency information. Based on the latency information, a route for a future request is selected, resources are allocated, or a user interface is presented. The process of determining the latency for content delivered by different routes may be repeated for content of different sizes. A future request is routed along the network path that provides the lowest latency for the data being requested.

Abstract: In various exemplary embodiments, a system and an associated method to retrieve information related to marketplace items within an electronic environment are disclosed. A network architecture comprises a listing module to store the information related to the marketplace items. The related information includes one or more user-created items such as an image of a model with one or more tagged items worn by the model. Each of the one or more tagged items has associated descriptive metadata. A communications module is arranged to receive a query including search terms related to social fashion items from an end-user. A query engine is coupled to the communications module to match the search terms contained within the query to the associated descriptive metadata of the tagged items in the listing module. A processing module then displays to the end-user the model and the one or more tagged items.

Abstract: A screen is presented on a display of a user device, which includes an element occupying less than an entirety of the screen. A second screen is then presented on the display, which includes a second element occupying less than an entirety of the second screen. The user device receives an input that requests that a previously presented element be presented within the second screen next to enable comparison of the previously presented element with the second element. The user device presents, on the display of the user device and responsive to receiving the input, a contemporaneous view of the element next to the second element within the second screen.

Abstract: A database node selects one of a number of database mirror hosts to perform a database read query. In theory, mirror hosts store redundant database entries. In practice, some mirror hosts fail to receive or retain some entries, e.g. if a mirror host was down for maintenance, a mirror host was overburdened, etc. The health of a mirror host, quantified as a health score, represents a likelihood that an insert statement was received and retained. Health scores are generated based on health metrics—data periodically retrieved from mirror hosts used to infer how well the database was operating, e.g. a count of entries that were inserted, a count of errors encountered while the entries were inserted, etc. If the database read query specifies entries inserted during a specific time period, the mirror host may be selected based on the health scores representing that time period.

Abstract: Systems and methods for multi-signal fault analysis are described. The system receives signal message information, over a network, from a collection device comprised of a plurality of mobile devices including a first mobile device and a second mobile device that are associated with a first user. The first signal message information includes a first maintenance message including characterization information that was received by the first mobile device from a component that includes a sensor that operates to sense a first part that is assembled into a vehicle. The second signal message information includes signal information that was received by the second mobile device including an audio signal that a microphone in the second mobile device sensed. The system analyzes the multi-signal information to diagnose a problem and communicate a message to the first user with a diagnosis of the problem.

Abstract: Methods, systems, and apparatus for facilitating a comparison of content in a user interface are described. Two or more open browser interfaces of a browser are identified, each of the two or more open browser interfaces containing an item description of one or more items. The browser interfaces include at least one of a tab of the browser or a window of the browser. One or more features related to each of the one or more items from at least two of the item descriptions are extracted or identified. The comparison for a plurality of items is generated based on the extracted or identified one or more features. The comparison of the plurality of items is displayed within at least one of the browser interfaces.

Abstract: Automatic workflow generation is described. One or more files containing code statements for accessing and modifying information in a destination database is received. The code statements are parsed from the one or more files and dependencies between the code statements are determined. A dependency graph is built by arranging the code statements according to the dependencies between the code statements. The dependency graph is partitioned by identifying at least one barrier code statement having an unclear dependency and dividing the dependency graph between code statements occurring prior to the at least one barrier code statement and code statements occurring after the at least one barrier code statement. Jobs are scheduled based on the partitioned dependency graph, and the code statements are annotated according to the scheduled jobs. A workflow is then automatically generated based on the annotated code statements.

Abstract: Methods and systems for displaying, to a user interface, categories of items for the user to select from for querying the system, in order to enable the system to return to the user images of items of the type desired by the user for possible purchase. The items categories of items can be presented to the user interface visually in silhouette form so that the user can select brand and the silhouette image of the type of item desired. Upon selection of the silhouette image, a query is sent to the system and images of the desired type of item are returned to the user interface for presentation to the user, along with certain attributes of the items.

Abstract: A system and method of identifying suspicious item-related features are disclosed. In some embodiments, a new item listing is received. The item listing may correspond to a request to list an item for sale on an e-commerce website. Item-related data for the item listing may be extracted. The item-related data can be used by a model to classify the new item listing. The model may be trained on data comprising identifications of which item listings in the sample of item listings are suspicious and identifications of which item listings in the sample of item listings are not suspicious may be received.

Abstract: In some examples, a system comprises a memory device for storing instructions and a processor which executes instructions causing the system to perform operations comprising receiving an instruction to transfer a state of a first device to a second device, and packaging information relating to the state of the first device in a file. The packaging of the information relating to the state of the first device includes recording each application executing on the first device in a list maintained in the file, and transferring the file containing information relating to the state of the first device to the second device, either directly or indirectly based on an availability of connections between the first device and the second device. The file, when processed by the second device, causes the second device to reproduce the state of the first device. In some example, reproducing the state of the first device includes the second device downloading, from one or more of the locations, one or more of the applications.

Abstract: Methods, systems, and computer programs are presented for adding new features to a network service. A method includes receiving an image depicting an object of interest. A category set is determined for the object of interest and an image signature is generated for the image. Using the category set and the image signature, the method identifies a set of publications within a publication database and assigns a rank to each publication. The method causes presentation of the ranked list of publications at a computing device from which the image was received.

Abstract: A machine may be configured to perform A/B testing on mobile applications. For example, the machine receives an identifier of a user from a mobile device that requests a layout description of a user interface, determines the layout description that provides a configuration of one or more elements of the user interface according to an experiment variant associated with an experiment, and causes display of the one or more elements of the user interface on the mobile device based on the layout description.

Abstract: An online experiment system is described that generates a randomization evaluation for an online experiment, while the experiment is ongoing, indicating whether a distribution of experiment participants allocated to one or more participant groups satisfies an expected distribution. The online experiment system analyzes one of the experiment groups to obtain an observed distribution of the subset of experiment participants included in the experiment group. The online experiment system then evaluates the observed distribution relative to the expected distribution for the experiment according to a decision criteria of a population stability index test. The decision criteria is influenced by a tuning parameter that represents a ratio of experiment participants included in the observed experiment group to experiment participants included in a different experiment group.

Abstract: Automatic workflow generation is described. One or more files containing code statements for accessing and modifying information in a destination database is received. The code statements are parsed from the one or more files and dependencies between the code statements are determined. A dependency graph is built by arranging the code statements according to the dependencies between the code statements. The dependency graph is partitioned by identifying at least one barrier code statement having an unclear dependency and dividing the dependency graph between code statements occurring prior to the at least one barrier code statement and code statements occurring after the at least one barrier code statement. Jobs are scheduled based on the partitioned dependency graph, and the code statements are annotated according to the scheduled jobs. A workflow is then automatically generated based on the annotated code statements.

Abstract: A rack mountable bracket is described. The bracket is mountable to a rack configured for mounting equipment. The rack is also configured according to a rack standard (e.g., EIA-310) that specifies how rails of the rack are disposed, one rail to another, and how a strip of holes disposed along a length of each rail is spaced, one hole to another. In one or more implementations, the bracket includes a pair of mounting edges for mounting the bracket to a mounting pair of the rails using hole strips of the mounting pair of rails. The bracket also includes an equipment mounting rail having a hole strip disposed along a length of the rail with holes spaced according to the rack standard. The equipment mounting rail is disposed between the mounting edges, such that when the bracket is mounted to the mounting pair of rails the bracket is horizontally oriented.

Abstract: Listing NFTs associated with physical items is described. Input specifying at least one obligation to be performed as a condition for transferring an NET from a first digital wallet to a second digital wallet is received. A listing for the NFT is then generated with information that describes the NET and the at least one obligation. A smart contract template is also generated, which includes instructions that are configured to ensure performance of the at least one obligation and transfer the NET from the first digital wallet to the second digital wallet. Responsive to purchase of the NFT via the listing, a smart contract is generated by updating the smart contract template with an identifier of the second digital wallet, and the NFT is transferred by executing the smart contract using a distributed state machine implemented on a blockchain.

Abstract: Methods and systems for improving memory utilization and access speed of database indexes are disclosed. A representation of attributes of items listed on a network-based publication system is identified. The representation includes a plurality of lists of the attributes as a sequence of serialized attributes, each of which stores an attribute identifier (ID) and data corresponding to the attribute identifier. A compact or compressed structure is used to represent the attribute IDs. This representation allows a reader application to quickly determine whether an attribute ID is present in the lists or not. Directly-addressable variable-length codes (DACs) are used, in an unconventional manner, to represent the data. By representing the data in this way, the index takes up less memory for certain types of data and the reader application is able to gain faster access to the index.

Abstract: Techniques for generating a digital wardrobe are presented herein. A transceiver can be configured to receive a request having a garment identifier and a user identifier. Additionally, an access module can be configured to access a first garment model, access a body model of the user corresponding to the user identifier, and access a second garment model corresponding to the user identifier. Furthermore, a processor can be configured by a garment simulation module to position the body model inside the first garment model and the second garment model, and calculate simulated forces based on the positioning. Moreover, a rendering module can be configured to generate an image of the garment models draped on the body model based on the calculated simulated forces. Subsequently, a display module can be configured to cause presentation of the generated image on a display of a device.

Abstract: A machine is configured to determine fashion preferences of users and to provide item recommendations based on the fashion preferences. For example, the machine accesses an indication of a fashion style of a user. The fashion style is determined based on automatically captured data pertaining to the user. The machine identifies, based on the fashion style, one or more fashion items from an inventory of fashion items. The machine generates one or more selectable user interface elements for inclusion in a user interface. The one or more user interface elements correspond to the one or more fashion items. The machine causes generation and display of the user interface that includes the one or more selectable user interface elements. A selection of a selectable user interface element results in display of a combination of an image of a particular fashion item and an image of an item worn by the user.

Abstract: Techniques are described, as implemented by computing devices, to control access to digital content through use of nonfungible tokens (NFTs). This is performed by leveraging a blockchain such that digital content associated with an item is made available to supplement use of the item (e.g., to supplement use of a physical item, digital content, and so forth) or make other functionality available based on a user's possession of the item.

Abstract: A system and method for obtaining relevant information for identifying items that are being listed for sale via electronic marketplace applications. More specifically, the systems and methods relate to obtaining relevant attribute values and/or information for a plurality of items at relatively the same time. In aspects, the systems and methods involve receiving an electronic file or folder having a plurality of electronic representations of each of the items. The electronic file or folder is then parsed to separate the different items for identification. Once separated, the system and methods involve sending the separate items to an identification service and retrieving the separate information for each item and returned to the requesting device.

Abstract: An error reporting system is described that generates, for a user interface being output by a computing device, a report using one or more test scripts for the user interface locally at the computing device. In response to detecting input at an error reporting control included as part of the user interface, the error reporting system obtains test scripts designated for testing the user interface and generates test results by executing the test scripts at the computing device outputting the user interface. Screen displays output by the computing device are recorded during execution of the test scripts to capture a visual appearance of the user interface and are aggregated together with the test results and feedback from a user of the computing device to generate the report. The report is then transmitted to a service provider associated with the user interface.

Abstract: Systems and methods change a user interface for the purpose of guiding a user in supplementing a product listing with an image to evidence the product's authenticity. Example embodiments include a machine-implemented method for accessing at least one database to retrieve an authenticity criterion mapped to a product and at least one reference image that depicts adequate detail of a product specimen to fulfill the authenticity criterion. The machine can further cause a user device to display the reference image to the user along with a suggestion that the user submit a candidate image depicting similar detail of the product. In some example embodiments, the method further includes retrieving the candidate image, confirming receipt of the candidate image, and displaying the candidate image, as well as adjusting a rank for a candidate specimen based on various factors.

Abstract: Systems and methods for multi-signal fault analysis are described. The system receives signal message information, over a network, from a collection device comprised of a plurality of mobile devices including a first mobile device and a second mobile device that are associated with a first user. The first signal message information includes a first maintenance message including characterization information that was received by the first mobile device from a component that includes a sensor that operates to sense a first part that is assembled into a vehicle. The second signal message information includes signal information that was received by the second mobile device including an audio signal that a microphone in the second mobile device sensed. The system analyzes the multi-signal information to diagnose a problem and communicate a message to the first user with a diagnosis of the problem.

Abstract: In various example embodiments, a system and method for computing a query feature score that can be used by a machine learning algorithm to rank search results is described. A query is received. The assigned probabilities for the arbitrary query is retrieved. The assigned probabilities is based on a probability model for a query feature. A score is computed for the query feature based on the assigned probabilities for the arbitrary query.

Abstract: A system and method of an intelligent shopping infrastructure is provided. A data storage of current inventory from a plurality of merchants is maintained. User preferences that indicate an item affinity of a user are retrieved. Furthermore, a location of the user is detected. Using the detected location, available inventory, and item affinity, an available item from the current inventory of a merchant located within a predetermined distance to the location of the user that match the item affinity of the user is identified. The available item is presented on a user interface of a client device of the user.

Abstract: A system is configured to detect a small, but meaningful, anomaly within one or more metrics associated with a platform. The system displays visuals of the metrics so that a user monitoring the platform can effectively notice a problem associated with the anomaly and take appropriate action to remediate the problem. A first visual includes a radar-based visual that renders an object representing data for a set of metrics being monitored. A second visual includes a tree map visual that includes sections where each section is associated with an attribute used to compose the set of metrics. Via the display of the visuals, the techniques provide an improved way of representing a large number of metrics (e.g., hundreds, thousands, etc.) being monitored for a platform. Moreover, the techniques are configured to expose useful information associated with the platform in a manner that can be effectively interpreted by a user.

Abstract: Example embodiments of the present disclosure include a system comprising a computer-readable storage medium storing at least one program and a computer-implemented method for providing a dynamic content delivery search system. In example embodiments, entry of a first keyword of a search string comprising a plurality of keywords is detected. The first keyword is identified after entry of the first keyword and prior to entry of a second keyword of the search string. A first set of matching results that match the identified first keyword is determined. Entry of the second keyword is detected and the second keyword identified. Using the second keyword, the first set of matching results is refined. The resulting search results are presented to a user.

Abstract: In an example, a text sentence comprising a plurality of words is obtained. Each of the plurality of words is passed through a deep compositional character-to-word model to encode character-level information of each of the plurality of words into a character-to-word expression. The character-to-word expressions are combined with pre-trained word embeddings. The combined character-to-word expressions and pre-trained word embeddings are fed into one or more bidirectional long short-term memories to learn contextual information for each of the plurality of words. Then, sequential conditional random fields are applied to the contextual information for each of the plurality of words.

Abstract: Technologies are shown for securing private information during shipping of an item storing shipping information at a blockchain address that includes a recipient geolocation address and generating a shipping label that includes a machine-readable pointer to the blockchain address. The label can be scanned to obtain the blockchain address and request the shipping information. In response to a request, the recipient geolocation address or a next intermediate shipper can be provided. The recipient can be notified to obtain authorization to release the recipient geolocation address or provide an alternative geolocation for delivery.

Abstract: Event data for an application execution is accessed from a table of logged events, the event data comprising a sequence, a hierarchy, and a start time and duration for each event. Dependency data for each event is also accessed to determine whether the start time for an event is dependent on the prior completion of at least one other event. A waterfall representation is then generated, the representation including an entry for each event in the sequence, with a start time and duration represented for each event. Based on the dependencies and hierarchy, it is determined, for each event with a start time that is later than the start time of an event which precedes it in the sequence, whether the event's start time is dependent on the prior completion of at least one preceding event. The start time for each event may then be advanced based on the determination.

Abstract: Aspects of the present disclosure involve an apparatus, systems, and methods for providing intelligent laundering alerts. Example embodiments involve an intelligent garment tag that may be affixed to garments, and may store and provide information about the garment to which it is affixed. The information stored thereon describes various aspects of the garment including color, fabric, and laundering information. The intelligent garment tags further include an alert component for providing laundering alerts (e.g., a flashing or colored light, a warning noise, or a vibration) based on, for example, garment color compatibility or laundering device (e.g., washing machine or clothes dryer) compatibility.

Abstract: Systems and methods for statistical dynamic geofencing are discussed. For example, a method for statistical dynamic geofencing can include operations such as receiving a location-based campaign request, analyzing demographic data, determining a geofence parameter, generating a geofence, and distributing location-aware advertisements. The location-based campaign request can include a list of points of interest (POIs) and a target demographic parameter. The demographic data can include data from geographic areas surrounding at least a portion of the POIs as well as data related to the target demographic parameter. The geofence can be generated based at least in part on the geofence parameter, where the geofence parameter directly affects either the size or shape of the geofence. The location-aware advertisements are distributed to a plurality of mobile devices determined to be within the geofence.

Abstract: In an online marketplace, buyers and sellers engage in transactions. Some transactions are fraudulent. The online marketplace uses a fraud detection system that identifies fraudulent transactions before an account holder is defrauded. In some example embodiments of the systems and methods described herein, a fraud detection system uses information that is indirectly associated with a transaction to determine if the transaction is fraudulent. A user interface is provided that shows relationships between accounts and objects. In response to input received via the user interface, the fraud detection system performs an action on an account, an object, or a transaction.

Abstract: Techniques for integrating a product model into a user supplied image are presented herein. A method is disclosed that includes receiving an image of a user space from a user, receiving one or more definitions that describe the user space, retrieving, based on search terms provided by the user, one or more items available at a networked marketplace that fit within the definitions of the user space, and generating a composite image by overlaying an image of one of the items on the image of the user space. The method further scales the image of the item according to the definition of the user space and the physical dimensions of the item, and depicts virtual guidelines indicating a current scale of the image of the item.

Abstract: A video is provided to viewers using a web-based platform without restricted audio, such as a copyrighted soundtrack. To do so, a video comprising at least two audio layers is received. The audio layers can include separate and distinct audio layers or a mix of audio from separate sources. A restricted audio element is identified in a first audio layer and a speech element is identified in a second audio layer. A stitched text string can be generated by performing speech-to-text on both audio layers and removing the text corresponding to the restricted audio element of the second audio layer. When playing back the video, a portion of the video is muted based on the restricted audio element. A voice synthesizer is employed to generate audible sound during the muted portion using the stitched text string.

Abstract: Methods for enhancing or automating a review process of annotation tags for a set of tokens is described. A system may receive a list of tokens with associated tags for each token for a data set and may output any identified inconsistencies where a token is assigned at least two different tags. For example, instead of a human looking at each token individually or taking a sample set of the tags for review, the described techniques may look at all tokens with the associated tags in a set of data and may leverage reorganizing the tokens and associated tags to highlight errors to be fixed. Accordingly, the system may look across all tokens within an entire data set, while a review (e.g., by a human) of possible errors of the data set is limited to the highlighted errors flagged by the system.