Abstract: A document-oriented database system generates an optimal query execution plan for database queries on an untyped field included in a collection of documents. Each untyped field is configured to store one or more scalar data values. The system generates histograms for scalar data values stored in untyped fields. Responsive to receiving a query statement for fetching or updating data value stored in one of the untyped fields, the system assigns a cost to each operator associated with the query statement based on the histogram, compares the costs assigned to the operators, and generates a query execution plan based on the comparison, and query the database using the query execution plan.

Abstract: A content generator system receives a request to generate content for a target entity, and one or more keywords. The content generator system retrieves, for the target entity, a current stage identifier linking the target entity to a current stage within a multi-stage objective. The content generator system generates an input vector including the current stage identifier, a target stage identifier, a token embedding comprising the one or more keywords, and a position embedding for each of the one or more keywords, the target stage identifier associated with a target stage within the multi-stage objective different from the current stage. The content generator system generates output text content for the target entity by applying a generative transformer network to the input vector. The content generator system transmits the output text content to a computing device associated with the target entity.

Abstract: Systems and methods for preserving storage efficiency during restoration of data from the cloud are provided. In one embodiment, a CBMAP is maintained that maps cloud block numbers (CBNs) to respective corresponding block numbers of a volume of a data storage system in which previously restored data has been stored by a previously restored file. By making use of the CBMAP during the restoration process, storage of duplicate file data blocks on the volume may be avoided by sharing with a current file being restored a reference to the corresponding file data block previously stored on the volume and associated with the previously restored file. In addition to preserving storage efficiency, use of the CBMAP facilitates avoidance of repeated GET operations for data associated with CBNs previously retrieved from the cloud and stored to the volume, thereby reducing data access costs as well as latency of the restore operation.

Abstract: A bot automatically compiles transaction information for a consumer from merchants that sell products in various categories and merchants that sell products in only a single category. The bot categorizes the consumer's purchases from the merchants and determines percentages of consumer spending in each category. The bot may obtain a category for purchases from merchants selling a single category of items based on a merchant identifier such as a merchant code. For other merchants, the bot may obtain category information from the merchant or from an analysis of a category code, or a product or service description or identifier. The categories and percentages of the consumer's purchases may be provided by the bot to the consumer or to a third party. The bot may provide special offers and promotions to a consumer based on the compiled category information and percentage spending by the consumer by category.

Abstract: In one aspect, the present disclosure relates to a method for reducing fraud in computer networks, the method including receiving, from each of a plurality of user devices, a request to block an ad displayed within a web browser installed on the user device, the request comprising image data and a forwarding URL associated with the ad; storing crowdsourced ad blocking data based on the received requests to block ads; receiving a request for a list of blocked ads; generating a list of blocked ads based on analyzing the crowdsourced ad blocking data, wherein analyzing the crowdsourced ad blocking data comprises identifying ads blocked by at least a threshold number of users; and sending the list of blocked ads to a first user device, the first user device comprising a browser extension configured to prevent ads within the list of blocked ads from being rendered in a browser.

Abstract: A database system stores data in a bi-temporal index that maintains recordation time and a validity time for recorded values. The index for a given key may be append only, such that the stored data for the key doesn't require modification of previously-recorded data for given validity times. The bi-temporal index comprises a key-value data store that allows values to be recorded for a specific key, validity time, and recordation time. To retrieve a value, a request presents a query with a key, subject time, and observation time. Queries look for values with a validity time most recent to and before the queried subject time and a recordation time, associated with the validity time, most recent to and before the queried observation time. The queries prioritize the validity time over the recordation time and returns the value that satisfies the query request.

Abstract: A computer-implemented method for dynamically modifying at least one visualization, the method comprising: receiving data from at least one database, receiving configuration data from at least one user, based on the received data, generating at least one report that includes at least one visualization based on the received data and the configuration data, receiving at least one annotation from at least one additional user, the at least one annotation including one or more tokens, transmitting at least one search query to the at least one database, the at least one search query including at least one of the one or more tokens, in response to transmitting the at least one search query, receiving updated visualization data from the at least one database, and displaying the updated visualization on one or more user interfaces of one or more user devices.

Abstract: A method includes receiving, at a container registry, a container image including a set of files, determining that at least one file of the set of files is a duplicate of a previously stored container image file at the container registry. The method further includes storing the container image at the container registry with the at least one file of the plurality of files that is a duplicate of the previously stored container image file removed from the container image.

Abstract: A method for presenting an associated conflict block comprises obtaining a first conflict file comprising a plurality of first code blocks, wherein a second code block of the first code blocks comprises a first code snippet of a current version of the second code block, a second code snippet of a basic version of the second code block, and a third code snippet of a to-be-merged version of the second code block. A first instruction triggered on a user interface for activating a to-be-processed code block is received, wherein a first type of the to-be-processed code block is of an unresolved conflict block or a single-side change block. A fourth code snippet comprised in the to-be-processed code block and indexes of N associated code blocks of the to-be-processed code block is presented on the user interface.

Abstract: Systems and methods for text description-based generation of avatars for Artificial Intelligence (AI) characters are provided. An example method includes receiving a description of a face of an AI character, where the description is in a free-text format and where the face of the AI character is rendered by an AI character model in a virtual environment; acquiring at least one parameter of the AI character model, where the at least one parameter corresponds to at least one facial feature of the face of the AI character; analyzing the description to generate at least one value for the at least one parameter; and assigning the at least one value to the at least one parameter.

Abstract: A deep learning-based Java program internal annotation generation method and system is provided. The method includes acquiring items with a Stars number ranked in the top, and extracting corresponding internal annotations and method statement lists; obtaining an <annotation, target code> pair; selecting an annotation in a Verb-dobj form; obtaining a code context associated with a target code segment; preprocessing the annotation, the target code, and the context to obtain a triplet dataset; randomly dividing the constructed dataset into a training set, a validation set, and a test set, and constructing an encoder-decoder network at the same time; enabling the training set in division to be used for model training, performing evaluation on the validation set to obtain a model with the best effect on the validation set as a target model; and predicting data in the test set with the obtained target model to generate a predicted annotation.

Abstract: Techniques allow a computer to responsively search for graph shapes similar to a user-selected graph shape much faster. Data can be pre-processed and stored as vectors, along with an index. The index can be used to find similar vectors that represent graph shapes similar to a user-selected shape in a computationally efficient manner. Vectors of multiple resolutions can be used to anticipate different sizes of a graph that a user can select, and comparisons can be repeated and refined. When a satisfactorily small number of candidate vectors are determined, more computationally intensive distance calculations can be performed on data reconstructed from the vectors.

Abstract: A user device stores a local application including initial search records and an initial search function. Each initial search record includes an application link that opens an application page and search data that describes the application page. The initial search function is configured to select from the initial search records. The user device executes the local application to request and receive search update data from a remote system. The search update data includes updated search records and an updated search function. The user device updates the initial search records and the initial search function with the updated search records and the updated search function. The user device receives a search query from a requesting application, selects a set of updated search records based on the search query and the updated search function, and provides a response to the requesting application including application links from the set of updated search records.

Abstract: Aspects described herein may provide systems, methods, and device for facilitating language learning using videos. Subtitles may be displayed in a first, target language or a second, native language during display of the video. On a pause event, both the target language subtitle and the native language subtitle may be displayed simultaneously to facilitate understanding. While paused, a user may select an option to be provided with additional contextual information indicating usage and context associated with one or more words of the target language subtitle. The user may navigate through previous and next subtitles with additional contextual information while the video is paused. Other aspects may allow users to create auto-continuous video loops of definable duration, and may allow users to generate video segments by searching an entire database of subtitle text, and may allow users create, save, share, and search video loops.

Abstract: Systems and methods are directed to generating a synoptic video from one or more documents. The system receives an indication to generate the synoptic video and in response, accesses the one or more documents and user preferences of a user that are derived by the network system. The one or more documents are analyzed to determine relevant text content, whereby the analyzing includes tagging the relevant text content. Based on the user preferences, the system generates the synoptic video that summarizes the tagged relevant text content. The content and style of the synoptic video is based on a user style indicated by the user preferences. For example, the user preference can determine a depth of the summarized content, a type of audio or text used, an amount of images to provide, and an amount of summarized text to provide. The synoptic video can then be displayed to the user.

Abstract: In some embodiments, a computer system is described. In some embodiments, the computer system receives, from a client application, a first request to access a collection of applications associated with an operation, wherein the collection of applications is associated with at least one third-party entity. In some embodiments, the computer system receives a selection of an application associated with a third-party entity from the collection of applications. In some embodiments, in response to receiving the selection, the computer system identifies a set of data comprising an aggregation of data stored on a plurality of databases required to perform the operation associated with the application. In some embodiments, the computer system determines that the application has access to the aggregation of data. In some embodiments, the computer sends a request to perform the operation to the application. In some embodiments, the computer system receives a set of results based on the performance of the operation.

Abstract: An operations management system and related method obtains user activity information representing user interactions with the operations management system responsive to an incident, the incident belonging to a category of incidents. The method represents the user activity information as an itemset. The method further processes the itemset with a mining algorithm to identify one or more items of information frequently accessed for the incident. The method yet further associates the identified one or more items of information with the category of incidents.

Abstract: Disclosed are embodiments for providing a domain-specific visualization of message content. Unclassified messages are received for a sender and a real-time classifier is used to assign categories to the messages. User interactions with email can then be used to generate a ranked list of domain-specific senders. This ranked list of senders and classified emails can then be used to provide a domain-specific view to a user. Further features (e.g., aggregated content pages by sender, dynamic call-to-action buttons, message previews etc.) can then be built on top of the ranked senders and messages.

Abstract: Methods, systems, and apparatus for efficient image analysis. In some aspects, a system includes a camera configured to capture images, one or more environment sensors configured to detect movement of the camera, a data processing apparatus, and a memory storage apparatus in data communication with the data processing apparatus. The data processing apparatus can access, for each of a multitude of images captured by a mobile device camera, data indicative of movement of the camera at a time at which the camera captured the image. The data processing apparatus can also select, from the images, a particular image for analysis based on the data indicative of the movement of the camera for each image, analyze the particular image to recognize one or more objects depicted in the particular image, and present content related to the one or more recognized objects.

Abstract: An automated system and method for assessing and rating accessibility are provided. A processor collects raw data corresponding to geographical objects obtained from sensors, the raw data including tagged and non-tagged data; operates on the raw data to extract features and reduce dimensionality of the raw data, thereby generating processed data having extracted features; generates accessibility data from the processed data; uses supervised machine learning techniques to develop models from the processed data; and implements the models and generates accessibility tags based on the extracted features. A database is configured to store geographical data related to the geographical objects and the accessibility tags corresponding to the geographical locations. An API is configured to access the database and provide a user interface for a user device to use an application to display the accessibility data and the accessibility tags on the user device customized to a disability.

Abstract: Systems and methods to detect attacks on the clocks of devices. In time sensitive networks are described. Particularly, the disclosed systems and methods provide detection and mitigation of timing synchronization attacks based on key performance indicators related to the protecting transmission windows in data streams of the time sensitive networks.

Abstract: Systems and methods for management of event metadata. The methods may include maintaining a plurality of data storage systems in communication with an external metadata management system. The methods may also include operating the metadata management system to store tagged event metadata corresponding to the plurality of data storage systems. Event metadata corresponding to a data storage system includes information associated with at least one data operation event executed on data residing in that data storage system.

Abstract: A method for executing a dynamically generated operation in a parallel processing framework may include executing the dynamically generated operation sequentially for a threshold quantity of times before initiating a compilation operation to generate precompiled code for the dynamically generated operation. Upon generating the precompiled code, execution of the precompiled code may be parallelized by determining, based on a quantity of time required to execute the precompiled code on a portion of a dataset, a task size. The execution of the precompiled code may be divided, based on the task size, into one or more tasks. Moreover, a corresponding quantity of worker threads may be allocated to perform the one or more tasks in parallel. Related methods and computer program products are also provided.

Abstract: Subject matter disclosed herein may relate to systems, devices and/or processes for computer implemented modeling of a multi-event process. In an embodiment, the method may include loading into one or more computer memory devices, a digitized representation of events of the multi-event process, in which a first one or more of the plurality of events is linked to a previous first one or more or of a second one or more of the plurality of events and generating a first visual representation of one or more visual indicators to represent progress of the first one or more of the plurality of events. The method may additionally include computing a respread progress profile of the first one or more events based, at least in part, on one or more modified attributes of the first one or more or of the second one or more of the plurality of events.

Abstract: Systems, methods, apparatuses, and computer program products for human or bot activity detection. The method may include, collecting time-series data on one or more events occurring on a webpage. The method may also include deriving classifications of the one or more events. The method may further include performing functional transformations of the time-series data. In addition, the method may include determining potential features of the one or more events based on a combination of the classifications of the one or more events, and results of the functional transformation. Further, the method may include training a machine learning model with the potential features. The method may also include determining, via the machine learning model, bot behavior and non-bot behavior of the one or more events.

Abstract: A method performed in real-time includes receiving and storing time-based data over a specific time period and dividing the specific time period into a plurality of time windows. The method further includes determining that data associated with two or more proximate time windows are within a predetermined variance of one another and responsive to the determination: generating a mathematical function representative of the data associated with the two or more proximate time windows, deleting the data associated with the two or more proximate time windows, and generating a representation of the deleted data from the mathematical function. In certain embodiments, the data comprises empirical network telemetry data.

Abstract: A content selection menu is generated and provided to a user. The content selection menu includes a plurality of audiovisual (AV) programs that are provided for user selection. The AV programs may be positioned in the content selection menu in accordance with their associated score for content attributes. Upon providing the content selection to a user, the user may select a program for viewing or the user may indicate interest in a program. If the user indicates interest in a program a second content selection menu is generated based at least in part on the attributes of the program of interest.

Abstract: Techniques for tracing attributes through queries in a set of nested queries are disclosed. A graphical user interface displays individual sub-queries as separate visual representations. Based on detecting a selection, a system analyzes data underlying the visual representations to identify queries, among the set of nested queries, that are associated with a particular attribute. One selection results in the system analyzing the underlying data to identify an origin query for an attribute, from among the set of nested queries. Another selection results in the system analyzing the underlying data to identify queries, from among the set of nested queries, that use the attribute received from a selected query. Another selection results in the system generating SQL code for a selected subset of queries, among the set of nested queries.

Abstract: Presented are systems and methods that construct BipartitE Graph INdices (BEGIN) embodiments for fast neural ranking. BEGIN embodiments comprise two types of nodes: sampled queries and base or searching objects. In one or more embodiments, edges connecting these nodes are constructed by using a neural network ranking measure. These embodiments extend traditional search-on-graph methods and lend themselves to fast neural ranking. Experimental results demonstrate the effectiveness and efficiency of such embodiments.

Abstract: Systems and methods for improving displays of media assets are disclosed herein. In an embodiment, a system receives a plurality of text comments from a plurality of devices to which a media asset was transmitted. The system analyzes the comments to identify text strings within the text comments. The system generates interactable elements from the text strings in the text comments, such that an interaction with the text string causes display of identifiers of media assets corresponding to the text string.

Abstract: Computer-implemented systems and methods for livestream engagement enhancement are disclosed and may be configured to retrieve a first set of texts; retrieve a second set of texts; transmit a plurality of texts including at least the first and second set of texts for display on one or more user interfaces associated with one or more user devices; receive user interaction data from at least one of the one or more user devices, wherein the user interaction data includes indication of user interaction by one or more users with at least one or more texts of the plurality of texts via at least one of the one or more user interfaces; and transmit each text of the one or more texts to a second user device for display on at least one of a first page or a second page of a second user interface.

Abstract: A communication system is implemented to provide multiple application services for a group of users, where each user of the group is able to access each of the multiple application services using a corresponding computing device. For each user of the group, the communication system provides the multiple application services by generating, on the user's computing device, a user-interface that includes multiple content views. The communication system operates to maintain one or more notification aggregation view for each user account, where individual users can view notifications generated by different application services and/or for different projects.

Abstract: An incongruent headline detection system receives a request to determine a headline incongruence score for an electronic document. The incongruent headline detection system determines the headline incongruence score for the electronic document by applying a machine learning model to the electronic document. Applying the machine learning model to the electronic document includes generating a graph representing a textual similarity between a headline of the electronic document and each of a plurality of paragraphs of the electronic document and determining the headline incongruence score using the graph. The incongruent headline detection system transmits, responsive to the request, the headline incongruence score for the electronic document.

Abstract: The present technology is effective to cause at least one processor to determine that a vehicle is within a threshold distance from a point of interest, determine details about the point of interest, and present the details about the point of interest. The vehicle may be characterized by having the ability to take any route to a destination requested by a passenger and to travel to the destination at any speed that is appropriate for road and traffic conditions. The details about the point of interest to present may be based upon a location and a speed of the vehicle.

Abstract: The systems and methods described herein relate to generative artificial intelligence systems using retrieval-augmented generation pipelines to supply information to large language models (LLMs). The potential for failures by such LLMs to return responses to prompts significantly increases with system complexity. To improve the resilience of the pipelines in handling such failures, various aspects described herein provide mechanisms for early detection and remediation of such prompt failure events. Thus, prompt failure events may be identified based upon (i) an elapsed time between sending a prompt and receiving a first token from the LLM exceeding a first threshold or (ii) an elapsed time between receiving such first token and receiving a last token exceeding a second threshold. Remediation may be achieved by causing a copy of the failed prompt to be sent to the LLM, without waiting for an error from the LLM provider or a standard network request timeout.

Abstract: Digital content search techniques are described. In one example, the techniques are incorporated as part of a multi-head self-attention module of a transformer using machine learning. A localized self-attention module, for instance, is incorporated as part of the multi-head self-attention module that applies local constraints to the sequence. This is performable in a variety of ways. In a first instance, a model-based local encoder is used, examples of which include a fixed-depth recurrent neural network (RNN) and a convolutional network. In a second instance, a masking-based local encoder is used, examples of which include use of a fixed window, Gaussian initialization, and an adaptive predictor.

Abstract: Systems, methods, and apparatus are provided for migrating data using an MR application. A smart glasses device may interface with a source database having a first schema. The smart glasses device may include an MR application. Using machine learning, the MR application may identify transfer data in a source database display for migration and convert a captured image of the transfer data to machine-encoded text. The smart glasses device may render the transfer data in the MR environment. The smart glasses device may connect to a target database having a second schema. Using machine learning, the MR application may derive the second schema and apply it to the transfer data. The MR environment may enable the user to modify the transfer data and replicate the modified data to generate a new sample set. The MR application may populate the target database with values from the new sample set.

Abstract: Systems and methods for pre-computing and memoizing simulations are disclosed. In an embodiment, a computer system may retrieve processing filters for running a simulation. The computer system may generate a simulation query based on the simulation filters and hash the simulation query. The computer system may determine whether a simulation result corresponding to the hashed simulation query has been pre-computed and stored. If the simulation result has been pre-computed and stored, the computer system may provide the simulation result in a user interface. If the simulation has not been pre-computed, the computer system may run a simulation and store the simulation results with a mapping to the hashed simulation query. If an identical simulation query is requested, the computer system may match a hash of the simulation query to a stored hash and provide the memoized simulation result without recomputing the simulation result, so simulation processing times can be reduced.

Abstract: Automatic rewriting of subtotal calculations including presenting, by a table manager, a spreadsheet structure on a GUI of a client computing system, wherein the spreadsheet structure presents a data set from a cloud-based data warehouse; receiving, at a first column of the spreadsheet structure by the table manager, a formula that includes a reference to a second column of the spreadsheet structure, wherein the formula is received from the client computing system; detecting, by the table manager, that the second column includes a subtotal calculation of a third column of the spreadsheet structure; and rewriting, by the table manager, the formula by replacing the reference to the second column with the subtotal calculation of the third column.

Abstract: Methods, computer systems, computer-storage media, and graphical user interfaces are provided for determining user affinities by tracking historical user interactions with tagged digital content and using the user affinities in content generation applications. Accordingly, the system may track user interactions with published digital content in order to generate user interaction reports whenever a user engages with the digital content. The system may aggregate the interaction reports to generate an affinity profile for a user or audience of users. A marketer may then generate digital content for a target user or audience of users and the system may process the digital content to generate a set of tags for the digital content. Based on the set of tags, the system may then evaluate the digital content in view of the affinity profile for the target user/audience to determine similarities or differences between the digital content and the affinity profile.

Abstract: A system for template invariant information extraction. The system comprises of processor, a first neural network model and a second neural network model. The processor is configured to recognize and extract entities and location of the entities in the input document using the first neural network model. The processor is further configured to classify whether the input document belongs to at least a template of the documents of the first training dataset using the second neural network model. The second neural network model comprises a linear classifier configured to generate a plurality of confidence scores for the input document corresponding to a unique template of the documents of the first training dataset. A threshold value to classify the input document belonging to template of the documents of the first training dataset is determined, Classification is done by comparing the confidence score with the threshold value.

Abstract: A central database system trains a machine-learned model based on training data identifying entity characteristics of account holder entities, content item characteristics of a content item presented to the account holder entities, and interactions between the account holder entities and the presented content item. The central database system then identifies a target set of account holder entities, and applies the trained machine-learned model to the entity characteristics of each account holder entity of the target set of account holder entities, the entity characteristics of each of the account holder entities that previously interacted with the content item, and the content item characteristics of the content item to identify a subset of the target set of account holder entities for presentation of the content item. The content item is then displayed to the subset, the content item includes an interface element that, when selected, causes an interaction to take place.

Abstract: Techniques for executing operations associated with sub-sets of queries in a set of nested queries are disclosed. A graphical user interface displays individual sub-queries as separate visual representations. Based on detecting a selection, a system executes a selected query in the set of nested queries, without executing parent queries in the set of nested queries. Based on detecting another selection, the system exports a selected query without exporting parent queries of the selected query. The system copies the metadata describing the selected query and transfers the metadata to another set of nested queries. Based on detecting another selection, the system displays a results count for a selected query. The system displays a number representing the number of results returned by the selected query, without displaying the query results.

Abstract: A contextual filter system configured to perform operations that include, capturing an image frame at a client device, wherein the image frame includes a depiction of an object, identifying an object category of the object based on the depiction of the object within the image frame, accessing media content associated with the object category within a media repository, generating a presentation of the media content, and causing display of the presentation of the media content within the image frame at the client device.

Abstract: A method may include presenting a user interface on a display device of a computing device, the user interface including: a search query input element; a plurality of graph type options; a graph level selection element; and a graph presentation area; receiving a search query inputted into the search query input element, the search query identifying a concept object in an ontology; retrieving data associated with the concept object from a graph database based on selections made in the graph type options and the graph level selection element, the data including a set of result objects related to the concept object; and rendering a hierarchical graph in the graph presentation area, the hierarchical graph illustrating the set of result objects and the concept object as interactive nodes.

Abstract: A method includes: obtaining, by a computing device, a first work item from a first index, wherein the first work item represents a namespace of a bucket of a vault in a dispersed storage network; dividing, by the computing device, the namespace into plural ranges of names; creating, by the computing device, plural second work items, each respective one of the plural second work items including a respective one of the plural ranges of names; and adding, by the computing device, each of the plural second work items to a second index.

Abstract: Implementations are provided for providing responsive audio recordings to user queries that are prerecorded by human beings, rather than generated automatically using speech synthesis processing. In various implementations, a query provided by a user at an input component of a computing device may be used to search a corpus of voice recordings From the searching, a plurality of candidate responsive voice recordings may be identified and ranked based on measures of credibility associated with speakers that created the candidate responsive voice recordings. Based on the ranking, one or more of the plurality of candidate responsive voice recordings may be provided for presentation to the user at an output component of the same computing device or a different computing device.

Abstract: An artificial intelligence (AI) method includes a processor receiving, from a human user, a query on an external database; performing transformations on the query to produce a refined query; generating an embedding of the refined query; and generating a system prompt and a user prompt based on the refined query. The processor then applies the embedding to a vector database by executing a similarity routine to identify discrete vectors in the vector database most similar to the embedding, and collects the most similar vectors for application to a large language model. Next, the processor applies the most similar vectors, the refined query, and the system and user prompts to a large language model to generate a comprehensive response to the query. The response includes a text document that is generated by execution of the large language model.

Abstract: Examples disclosed herein may involve a computing system that is operable to (i) receive one or more images related to a global map having a plurality of overlapping map segments, wherein each of the plurality of overlapping map segments overlaps with one or more neighboring map segments, (ii) based on a preliminary location determination for the one or more images, identify at least a first overlapping map segment of the plurality of overlapping map segments that corresponds to the one or more images, (iii) generate a reconstruction of the first identified overlapping map segment based on the one or more images, and (iv) fuse the generated reconstruction of the first identified overlapping map segment together with the first identified overlapping map segment's one or more neighboring map segments based on overlapping map portions between the generated reconstruction and the first identified overlapping map segment's one or more neighboring map segments.

Abstract: Indices of storage systems are managed. An example method includes: receiving a flush cycle for flushing expired events in a storage system including multiple events in a data stream, determining, based on the flush cycle, a time slice for managing the index of the storage system, creating a slice index node for the time slice in response to determining that the occurrence time of a first event that enters the storage system in the data stream is within the time slice, wherein the slice index node includes an index node of the first event, and adding the slice index node to the index. A corresponding device and a corresponding computer program product are provided. Thus, the index of a large number of events entering the storage system is managed according to the occurrence time of the events, and the storage system can be searched and updated accurately and effectively.