Abstract: In some examples, learning based incident or defect resolution, and test generation may include ascertaining historical log data that includes incident or defect log data associated with operation of a process, and generating, based on the historical log data, step action graphs. Based on grouping of the step action graphs with respect to different incident and defect tickets, an incident and defect action graph may be generated to further generate a machine learning model. Based on an analysis of the machine learning model with respect to a new incident or defect, an output that includes a sequence of actions may be generated to reproduce, for the new incident, steps that result in the new incident, reproduce, for the new defect, an error that results in the new defect, identify a root cause of the new incident or defect, and/or resolve the new incident or defect.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for creating structured data using data received from unstructured textual data sources. One of the methods includes receiving unstructured textual data, identifying one or more keywords in the unstructured textual data, determining one or more patterns included in the unstructured textual data using the identified keywords, identifying one or more intelligence types that correspond with the unstructured textual data using the determined patterns, and associating, for each of the identified intelligence types, a data subset from the unstructured textual data with the respective intelligence type.

Abstract: A system and method for extracting structured information from an implicit table is disclosed. The system and method provide a way to locate and categorize structured information from an implicit table. More specifically, the system and method provide a way of determining which part of an input image document includes a dominant table and which parts of the dominant table make up rows and columns. These details give meaning to the structured information of the implicit table. These details can be used to properly place the structured information from the implicit table into a two-dimensional data structure, such as a data structure in a relational database. In other words, the structured information from a scanned or digital Portable Document Format (PDF) document can be extracted and placed into a useful format, such as a relational database.

Abstract: A device may determine a behavioral pattern of an account over a past time period based on data relating to one or more transactions associated with the account. The device may identify one or more quantitative features of the behavioral pattern and one or more spatial features of the behavioral pattern. The device may determine an account type cluster to which the account belongs, based on the one or more quantitative features and the one or more spatial features identified. The device may determine, based on the account type cluster that is determined, a model for processing the behavioral pattern. The device may predict, using the model that is determined, an amount of funds that is likely to remain in the account during a future time period. The device may perform one or more actions based on the amount of funds that is predicted.

Abstract: A system and method for transforming unstructured text into structured form is disclosed. The system and method include converting an input word sequence (e.g., sentence) into tagged output which can be then easily be converted into a structured format. The system may include a bidirectional recurrent neural network that can generate multiple labels of individual words or phrases. In some embodiments, a customized learning loss equation involving set similarity is used to generate the multiple labels.

Abstract: A system and method for extracting structured information from image documents is disclosed. An input image document is obtained, and the input image document may be analyzed to determine a skeletal layout of information included in the input image document. A measure of similarity between the determined skeletal layout and each of the document templates may be determined. A document template may be selected as a matched template, based on the determined measure of similarity. Box areas from the input image document may be cropped out, and optical character recognition (OCR) may be performed on the box areas. Obtained recognized text may be automatically processed using directed search to correct errors made by the OCR. Statistical language modeling may be used to classify the input image document into a classification category, and the classified input image document may be processed according to the classification category.

Abstract: In some examples, artificial intelligence based music playlist curation may include ascertaining listening data for a plurality of tracks, and generating a plurality of embeddings that represent the plurality of tracks. A replacement track for an existing track in an input playlist may be generated. Alternatively or additionally, at least one additional track may be added to the input playlist. Alternatively or additionally, based on a seed set of tracks, an output playlist that includes a specified number of tracks that is greater than a number of tracks in the seed set of tracks may be generated. Alternatively or additionally, based on a plurality of specified attributes, the plurality of embeddings may be partitioned into a plurality of clusters corresponding to the plurality of specified attributes.

Abstract: In some examples, artificial intelligence based music playlist reordering and song performance assessment may include ascertaining listening data for a plurality of tracks, and generating a plurality of sessions. Tracks that have been played more than a specified play threshold may be identified and retained. Sessions that are greater than a minimum session length threshold and less than a maximum session length threshold may be retained. Input-output track sequences may be generated for the retained sessions. Unique identifiers may be assigned to each of the tracks present across the retained sessions. Each input-output track sequence may be vectorized based on associated unique identifiers. A neural network model may be trained based on the vectorized input-output track sequences. For an input playlist, the trained neural network model may be used to generate a modified playlist. Additionally or alternatively, a user-song interaction graph may be used to generate another modified playlist.

Abstract: A system and method for training a virtual agent to identify a user's intent from a conversation is disclosed. The system and method use an iterative process of clustering multiple conversations (converted into feature representations) used for training a machine learning model into labeled clusters having similar user intents. Clustering enables labeling a large number of training conversations efficiently. The labeled clusters may be used to train a virtual agent to classify the conversational intent of a conversation. Then, the machine learning model can classify future conversations based on similarity to labeled clusters. By knowing a human user's intent, a virtual agent can deliver what the user desires.

Abstract: In some examples, learning based incident or defect resolution, and test generation may include ascertaining historical log data that includes incident or defect log data associated with operation of a process, and generating, based on the historical log data, step action graphs. Based on grouping of the step action graphs with respect to different incident and defect tickets, an incident and defect action graph may be generated to further generate a machine learning model. Based on an analysis of the machine learning model with respect to a new incident or defect, an output that includes a sequence of actions may be generated to reproduce, for the new incident, steps that result in the new incident, reproduce, for the new defect, an error that results in the new defect, identify a root cause of the new incident or defect, and/or resolve the new incident or defect.

Abstract: Methods and systems including computer programs encoded on a computer storage medium, for interactive content recommendation. In one aspect, a method includes receiving a request for content by a user, determining a user intent based on the received request, providing to the user a first attribute responsive to the user intent, receiving a first attribute value responsive to the first attribute, providing a second attribute, and receiving a second attribute value responsive to the second attribute. A particular content vector including a first content attribute and a second content attribute for a particular content item is identified where the first content attribute and the second content attribute sufficiently match the first attribute value and the second attribute value. The particular content item is provided as a suggested content item, and, responsive to a user selection of the particular content item, provided for presentation on the user device.

Abstract: A device may receive information associated with first and second bug reports to be classified as duplicate or non-duplicate bug reports. The device may identify first and second descriptions associated with the first and second bug reports, respectively. The first and second descriptions may be different descriptions having a shared description type. The device may identify a neural network for encoding the first and second descriptions, based on the shared description type. The device may encode the first description into a first vector using the neural network, and may encode the second description into a second vector using the neural network. The device may classify the first and second bug reports as duplicate or non-duplicate bug reports based on the first vector and the second vector. The device may perform an action based on classifying the first and second bug reports as duplicate or non-duplicate bug reports.

Abstract: A system and method for spoken language understanding using recurrent neural networks (“RNNs”) is disclosed. The system and method jointly performs the following three functions when processing the word sequence of a user utterance: (1) classify a user's speech act into a dialogue act category, (2) identify a user's intent, and (3) extract semantic constituents from the word sequence. The system and method includes using a bidirectional RNN to convert a word sequence into a hidden state representation. By providing two different orderings of the word sequence, the bidirectional nature of the RNN improves the accuracy of performing the above-mentioned three functions. The system and method includes performing the three functions jointly. The system and method uses attention, which improves the efficiency and accuracy of the spoken language understanding system by focusing on certain parts of a word sequence. The three functions can be jointly trained, which increases efficiency.

Abstract: In the pharmaceutical research and development process, it may be necessary to process large amounts of medical records or clinical literature, to ensure safety of patients consuming a drug. A pharmacovigilance system may assist in this process by efficiently and automatically processing medical records to extract information and relationships contained therein and may also form a preliminary assessment regarding a medical or clinical judgment. The pharmacovigilance system may automatically generate reports based on this information, which may be validated by trained clinicians and medical experts.

Abstract: A device may receive, from a user device, an implicit search request identifying a resource accessed by a user of the user device. Based on properties defined in a user role and a rule identified in context configuration information, the device may determine an intent associated with the user and a context associated with the intent and the resource accessed by the user. The device may send, based on query options identified in search configuration information, a search query identifying search criteria associated with the context, and may receive a search response identifying available resources that satisfy the search criteria. The device may determine a probability, for each of the available resources, that the available resource matches the context, identify a portion of the search response based on the probabilities, and send the portion to the user device.

Abstract: A rapid application developer system is disclosed that provides a development platform configured to generate an application manifest document for developing a new target application within a certain industry. The application manifest is generated based on components identified from previously developed applications within the same industry as the target application.

Abstract: A device receives a corpus of text documents, and utilizes feature extraction on a text document, of the corpus of text documents, to generate features from the text document, where the features include binary features, numeric features, and categorical features. The device performs feature engineering on one or more of the binary features, the numeric features, or the categorical features, to generate converted features, and performs feature encoding on the text document, based on the converted features, to represent the text document as a vector with a similarity score for a domain. The device provides the vector with the similarity score for the domain, as training data, to a machine learning model to generate a trained machine learning model, and performs an action using the trained machine learning model.

Abstract: A system and method for extracting structured information from an implicit table is disclosed. The system and method provide a way to locate and categorize structured information from an implicit table. More specifically, the system and method provide a way of determining which part of an input image document includes a dominant table and which parts of the dominant table make up rows and columns. These details give meaning to the structured information of the implicit table. These details can be used to properly place the structured information from the implicit table into a two-dimensional data structure, such as a data structure in a relational database. In other words, the structured information from a scanned or digital Portable Document Format (PDF) document can be extracted and placed into a useful format, such as a relational database.

Abstract: A system and method for extracting structured information from image documents is disclosed. An input image document is obtained, and the input image document may be analyzed to determine a skeletal layout of information included in the input image document. A measure of similarity between the determined skeletal layout and each of the document templates may be determined. A document template may be selected as a matched template, based on the determined measure of similarity. Box areas from the input image document may be cropped out, and optical character recognition (OCR) may be performed on the box areas. Obtained recognized text may be automatically processed using directed search to correct errors made by the OCR. Statistical language modeling may be used to classify the input image document into a classification category, and the classified input image document may be processed according to the classification category.

Abstract: The system and method generally include identifying whether an utterance spoken by a user (e.g., customer) is a complete or incomplete sentence. For example, the system may include a partial utterance detection module that determines whether an utterance spoken by a user is a partial utterance. The detection process may include providing a detection advice code that gives a recommendation for handling the utterance of interest. If it is determined that the utterance is an incomplete sentence, then the system and method can identify the type of utterance. For example, the system may include a partial utterance classification module that predicts the class of a partial utterance. The classification process may include providing a classification advice code that gives a recommendation for handling the utterance of interest.