Abstract: Techniques are provided for improving computers as tools for assisting in negotiations. Specifically, techniques are provided for using a trained machine learning system to predict the likelihood that a party to a negotiation intends to comply with terms that are under consideration. In some negotiations, each party of a negotiation may use the techniques described herein to determine terms to offer the other party. In such situations, both parties may be both terms-receiving parties and terms-offering parties. By using a trained machine learning system to predict the intent of a party, the possibility of human bias significantly reduced, allowing proposed terms to be based more on objective facts and predictive indicators rather than the prejudices of the agents that have been delegated the responsibility of proposing terms.

Abstract: Embodiments collect, from an event stream having both client-side and server-side information from multiple tracking sources, user interaction records for a web application and supporting systems. The records are stored in a data store, and are retroactively correlated with conclusive user identifiers and, if applicable, with areas of interest. Attribution of interaction records to conclusive identifiers and areas of interest is based on links that have been established between identifiers. Established identifier links may be subject to one or more link removal criteria. If link removal criteria is satisfied for an established link, the link is broken. Once a link is broken, interaction records, found in the data store, that are associated with the inconclusive identifier of the broken link are no longer associated with the conclusive identifier of the broken link. Each interaction record that is associated with a conclusive identifier is assigned a confidence metric for the association.

Abstract: Techniques a provided for performing multi-system operations in which changes are asynchronously committed in multiple systems. Metadata about the multi-system operation is injected into the commit logs of one system involved in a multi-system operation. An event stream is generated based on the commit logs of the one system, and is used to drive the operations that one or more other systems need to perform as part of the multi-system operation. A reconciliation system reads the logs of all systems involved in the multi-system operation and determines whether the multi-system operation completed successfully. Techniques are also provided for using machine learning to generate models of normal execution of different types of operations, detect anomalies, pre-emptively send expectation messages, and automatically suggest and/or apply fixes.

Abstract: Techniques a provided for performing multi-system operations in which changes are asynchronously committed in multiple systems. Metadata about the multi-system operation is injected into the commit logs of one system involved in a multi-system operation. An event stream is generated based on the commit logs of the one system, and is used to drive the operations that one or more other systems need to perform as part of the multi-system operation. A reconciliation system reads the logs of all systems involved in the multi-system operation and determines whether the multi-system operation completed successfully. Techniques are also provided for using machine learning to generate models of normal execution of different types of operations, detect anomalies, pre-emptively send expectation messages, and automatically suggest and/or apply fixes.

Abstract: Techniques for simulating human interface interaction resilient to user interface changes are presented. Automated analysis is performed on an image of a user interface of a computer program that is to be a target of simulated user interactions, and that determines the location and interaction purpose of a particular control on the user interface. An interaction package is read, and it indicates one or more interaction purposes that are to be carried out through interactions with the user interface of the computer program. Interaction with the user interface is simulated to carry out the one or more interaction purposes, where the particular interaction purpose is among the one of the one or more interaction purposes and simulating user interactions includes simulating user interaction with the particular control, based on the determined location and particular interaction purpose.

Abstract: Embodiments collect, from an event stream having both client-side and server-side information from multiple tracking sources, user interaction records for a web application and supporting systems. The records are stored in a data store, and are retroactively correlated with conclusive user identifiers and, if applicable, with areas of interest. Attribution of interaction records to conclusive identifiers and areas of interest is based on links that have been established between identifiers. Established identifier links may be subject to one or more link removal criteria. If link removal criteria is satisfied for an established link, the link is broken. Once a link is broken, interaction records, found in the data store, that are associated with the inconclusive identifier of the broken link are no longer associated with the conclusive identifier of the broken link. Each interaction record that is associated with a conclusive identifier is assigned a confidence metric for the association.

Abstract: An item is encrypted to create a unique hash-value for the item. From this hash-value, an item can be uniquely identified. The hash-value for the item is stored in a first blockchain. When the item is included in a transaction, a transaction entry is stored in a block of the first blockchain. When an item participates in a group of items, a group of items is formed. The group of items is encrypted to create a unique hash-value for the group of items. The hash-value for the group of items may be based on hash-values from the items included in the group of items. The hash-value for the group of items is stored in a second blockchain that is distinct from the first blockchain. When the group of items is included in a transaction, a transaction entry is stored in a block of a second blockchain.

Abstract: Embodiments provide a batching system that conforms message batches to publication constraints and also to message ordering requirements. An output array of messages is formed from messages received from a plurality of input streams, in which the messages are ordered. The output array preserves the ordering of the messages found in the source input streams. Messages are added from a head of the output array to a batch until addition of a next message to the batch would violate a particular batch processing constraint imposed on the batch. According to embodiments, one or more additional messages are included in the current batch when addition of the one or more additional messages to the batch (a) does not violate the particular batch processing constraint, and (b) continues to preserve the ordering of the messages, in the batch, with respect to the respective ordering of each of the plurality of input streams.

Abstract: Techniques for automatically attaching optical character recognition data to images are provided. The techniques include receiving an image file containing an image and performing optical character recognition on the image to generate text output. The techniques then continue by identifying a particular text item from within the generated text output and determining that the particular text item is a value for particular corresponding key. Then metadata that indicates that the particular text item is a value for the particular key is stored in the image file.

Abstract: Techniques are provided for using tokenization in conjunction with “behind-the-wall” JWT authentication. “Behind-the-wall” JWT authentication refers to JWT authentication techniques in which the JWT stays exclusively within the private network that is controlled by the web application provider. Because the JWT stays within the private network, the security risk posed by posting the JWT in a client cookie is avoided. However, because JWT is used behind-the-wall to authenticate a user with the services requested by the user, the authentication-related overhead is significantly reduced.

Abstract: A queue-based task management system is provided. Unlike conventional queue-based task management systems, the system described herein does not handle all tasks in the conventional manner. Rather, tasks can be associated with one of several modes, including: a queue-mode (always storing the task in the queue) and an immediate-mode (record the task, but execute it immediately if possible). The two modes may be controlled programmatically to optimize utilization of system resources. Immediate-mode is implemented by monitoring available task system resources, and executing the task immediately in-process if there are resources available; otherwise the task is delegated to the queue. This in-process execution allows the sharing of parent task resources.

Abstract: Embodiments facilitate efficient recovery from an inter-stream messaging system failure by using heartbeat messages (HMs) to act as watermarks for message recovery. Embodiments insert HMs into each of the input streams at configurable regular intervals. The inter-stream router determines that a message being routed is a HM, and corresponding output HMs are generated based on the input HM. Embodiments insert a respective output HM into each of the output streams. Information indicating which output HMs have been processed from the output streams is tracked. After a failure of the inter-stream router, embodiments identify a target HM for each input stream, which is the latest HM sourced from the respective input stream that was processed from all of the output streams. After the inter-stream router restarts, the router initiates message routing, from each input stream, at the location of the respective input stream's target HM within the input stream.

Abstract: Techniques a provided for communication graph tracking in multi system operation in heterogeneous database systems, including retrieving two or more database commit logs from two or more corresponding heterogeneous databases. For each of the commit logs, a subset of information is extracted, including a correlation identifier, and event stream entries are produced based on that information and are stored. The event stream entries are then retrieved and event database entries are created based on the event stream entries. A request for a communication graph related to a particular correlation identifier is received and the event database is searched for matching entries. Those entries are used to generate a displayable communication graph for the particular correlation identifier, which is then provided in response to the request.

Abstract: Techniques for optical character recognition improvement based on inline location determination is provided. The techniques include receiving a digital data stream containing a digital image. As the digital data stream arrives, a determination is made whether a number of received bytes associated with a header portion of the digital image has reached a target number. In response to determining that the number of received bytes associated with the header portion of the digital image has reached the target number, the bytes associated with the header portion of the digital image are cloned. While the digital data stream is received, location data from the cloned bytes associated with the header portion are determined. After the digital image has been received, text in the digital image is caused to be recognized by an optical character recognition system based, at least in part, on the location data.

Abstract: A bulk-selection control is described in which the same user input action automatically causes both (a) scrolling of a list of visual indicators of items, and (b) selection of the items that correspond to the visual indicators. Specifically, as the visual indicator of an item is scrolled beyond a “selection-triggering position”, the visual indicator is visually marked to indicate that the corresponding item is selected, and the corresponding item is automatically added to a set of selected items. When scrolled the opposite direction, the items that have already been added to the set of selected items are not removed therefrom, even if the scrolling operation causes their visual indicators to cease to be beyond the selection-triggering position.

Abstract: Embodiments provide a batching system that conforms message batches to publication constraints and also to message ordering requirements. An output array of messages is formed from messages received from a plurality of input streams, in which the messages are ordered. The output array preserves the ordering of the messages found in the source input streams. Messages are added from a head of the output array to a batch until addition of a next message to the batch would violate a particular batch processing constraint imposed on the batch. According to embodiments, one or more additional messages are included in the current batch when addition of the one or more additional messages to the batch (a) does not violate the particular batch processing constraint, and (b) continues to preserve the ordering of the messages, in the batch, with respect to the respective ordering of each of the plurality of input streams.

Abstract: User interface controls are provided for transferring an asset in an intuitive and unambiguous manner. According to one embodiment, the user interface has a “source indicator” that represents the source of asset transfer, and a “target indicator” that represents a potential recipient of an asset transfer. While the source and target indicators are visibly displayed on the display of a computing device, a user transfers the asset from the source represented by the source indicator to the target indicated by the target indicator by performing a dragging operation that drags the source indicator in the direction of the target indicator. The edge of the source indicator that faces the target indicator, and the edge of the target indicator that faces the source indicator, are interlocking. That is, rather than have linear edges, the respective facing edges have non-linear shapes so that, when dragged together, the shapes interlock.

Abstract: Techniques for state analysis for one or more virtual machines are provided herein. The techniques include receiving periodic thread dumps from each of one or more virtual machines and determining whether one or more actions should be taken based on the periodic thread dumps for at least one virtual machine of the one or more virtual machines. When it is determined that the one or more actions should be taken based on the periodic thread dumps for the at least one virtual machine of the one or more virtual machines, those the one or more actions associated with satisfying the one or more rules engines are performed. In some embodiments, the rules engines may include a trained supervised or unsupervised machine learning model.

Abstract: Techniques are provided for predicting time remaining for currently-execution jobs. Rather than predict time remaining based on prior executions, time remaining is predicted based on what has happened so far in the current execution. In order to generate predictions for a currently-executing job instance based on statistics about the currently-executing job instance, the system tracks, for each monitored job instance: (a) how many completed-tasks are currently associated with the job instance, and (b) how many created-but-not-completed tasks are associated with the job instance, and then predicts (c) how many not-yet-created tasks the job instance is likely to have.

Abstract: Embodiments utilize a unified testing framework (UTF) to automate browser-based API testing of web applications. Embodiments cause a browser to generate an authentication context for a user of the web application, and then test API functionality based on the established authentication context. The UTF is configured to run both UI testing and API testing. Furthermore, in order to facilitate running multiple API tests in a given testing session (potentially being performed in parallel), the UTF stores results of a given API test in a corresponding data island DOM node. Each data island node, generated for each respective API test being run in a given testing session, is identified by a unique global unique identifier (GUID) that is associated with the corresponding API test. Using this GUID, the UTF accesses the test results of each API test individually as the test results are asynchronously returned by the remote service being tested.