International Business Machines Corporation

Hima Patel

Nitin Kumar Gupta

Sameep Mehta

Ruhi Sharma Mittal

Shashank Mujumdar

Shanmukha Chaitanya Guttula

Kuntal Dey

Berthold Reinwald

Pedro Miguel Barbas

Pranay Kumar Lohia

Shaikh Shahriar Quader

Balaji Ganesan

Adrian Mahjour

Soma Shekar Naganna

Kalapriya Kannan

Akshar Kaul

Neeraj Ramkrishna Singh

Naveen Panwar

Anush Sankaran

Shettigar Parkala Srinivas

Vitobha Munigala

Jayachandu Bandlamudi

Lokesh Nagalapatti

Riddhiman Dasgupta

Akshay Parekh

Daivik Swarup Oggu Venkata

Methods, systems, and computer program products for automatically improving data annotations by processing annotation properties and user feedback are provided herein. A computer-implemented method includes obtaining data annotation pairs, each comprising an input data annotation in a first format and a corresponding output data annotation in a second format; determining, within at least a portion of the data annotation pairs, one or more non-diffs; identifying, across the at least a portion of data annotation pairs, data annotation properties associated with multiple intents by processing the non-diffs using property-related rules; modifying at least a portion of the data annotation pairs based on the identified data annotation properties; outputting the modified data annotation pairs to at least one user; and generating a final collection of data annotation pairs by processing at least a portion of the modified data annotation pairs and user feedback received in response to the outputting.

Automatically improving data annotations by processing annotation properties and user feedback

One embodiment provides a computer implemented method, including: obtaining an information document corresponding to an entity, wherein the information document includes redacted information spans; identifying an entity type for each of the redacted information spans, wherein the entity type identifies a relationship between a redacted information span and at least one other entity within the information document; replacing the redacted information spans with replacement entities corresponding to the entity type of a given redacted information span, wherein the replacing is performed in view of a frequency distribution of actual information and wherein the replacing includes maintaining relationships of the redacted information spans; and controlling bias within the replacement entities, wherein the controlling includes detecting bias within the replacement entities.

Generation of training data from redacted information

A method, computer program, and computer system are provided for collecting and annotating data based on user preference. Unlabeled data corresponding to one or more entries within a dataset is received. Pseudo-labeled data is generated based on the unlabeled data. Based on one or more quality metrics, each entry from among the pseudo-labeled data is determining to be included within a final dataset. A user is prompted for annotations corresponding to entries of the pseudo-labeled data included within the final dataset. A determination is made as to whether additional data is needed based on comparing the final dataset to the one or more quality metrics, and the additional information is collected if the final dataset does not meet the quality metrics.

Annotating and collecting data-centric AI quality metrics considering user preferences

A cleansing operation defined for a data structure of a database managed by a database management system is obtained. The cleansing operation is performed on data of the data structure to obtain clean data. The cleansing operation that is defined for the data structure and performed on data of the data structure is performed by the database management system.

In-database data cleansing

Selecting and ordering the execution of data quality rules includes generating a snapshot of a table-formatted dataset. The snapshot comprises a reduced number of rows of the dataset such that each column variation of the dataset is included in the snapshot. A predetermined collection of data quality (DQ) rules is executed on the snapshot. One or more performance statistics is determined for each of the DQ rules. The performance statistics indicate a likelihood that a DQ rule determines a data quality deficiency. Based on the performance statistics, a subset of the DQ rules is generated. Each DQ rule of the subset is selected based on the likelihood that the DQ rule selected detects a quality deficiency. An ordered subset of selected DQ rules is generated by ordering the application of each of the subset of DQ rules selected. The ordering specifies a sequence for executing each selected DQ rule.

Automated selection and ordering of data quality rules during data ingestion

A cleansing operation is performed on data of a data structure to obtain clean data. The clean data is stored as part of the data structure; however, the clean data is independent of the data. A mapping is performed to provide a set of ordered data that includes the data and the clean data.

In-database data cleansing and independent store of clean data

Processing within a computing environment is facilitated using a corpus processing system to assess and enhance quality of a corpus of unstructured documents for a specified task. The processing includes referencing, by a corpus processing engine, the corpus of unstructured documents to obtain unstructured document data, and applying, by a corpus quality metrics engine, a set of quality metrics to the document data to obtain a set of quality metric scores. Further, the process includes automatically selecting, by a quality metric selection engine, a subset of task-relevant quality metrics using the quality metric scores and the specified task, and automatically transforming, at least in part, multiple documents of the corpus to remediate one or more identified issues with the documents. The automatically transforming results in remediated documents tuned for the specified task, which are provided for the specified task to be performed.

Corpus quality processing for a specified task

Automatically selecting data for machine learning datasets is provided. The method comprises receiving an input dataset and user-specified data quality metrics. The input dataset is matched to a subset of candidate datasets in a repository according to schema characteristics. A second subset of candidate datasets having a distance from the input dataset above a specified threshold is selected from the first subset of candidate datasets. The second subset of candidate datasets are merged into a merged dataset. Top ranked samples above a specified second threshold are identified from the merged dataset based on the user-specified data quality metrics. The input dataset, augmented with the top ranked samples, is returned to the user.

Automatically selecting relevant data based on user specified data and machine learning characteristics for data integration

Embodiments are disclosed for a method. The method includes receiving an annotation set for a machine learning model. The annotation set includes multiple data points relevant to a task for the machine learning model. The method also includes determining total weights corresponding to the data points. The total weights are determined based on multiple ordering constraints indicating multiple data classes and corresponding weights. The corresponding weights represent a relative priority of the data classes with respect to each other. The method further includes generating an ordered annotation set from the annotation set. The ordered annotation set includes the data points in a sequence based on the determined total weights.

Ordering annotation sets for machine learning

A computer implemented method transforms data. Responsive to receiving a data transformation of an input string to an output string, a computer system identifies mappable tokens in the input string that are mappable to the output string. The computer system creates a set of initial mappings for a set of common tokens in the mappable tokens. The set of initial mappings maps the set of common tokens from the input string to the output string. The computer system creates a set of user mappings that maps the mappable tokens from input string to the output string using a user input to the set of initial mappings. The computer system generates program code that transform input strings to output strings using the set of user mappings that maps the mappable tokens from input string to the output string, wherein the program code is used to transform input strings to output strings.

Data transformation methodology using generated program code and token mappings

One embodiment provides a method, including: receiving a sample set for training a machine-learning model, wherein the sample set includes a plurality of classes, wherein classes within the plurality of classes have an imbalance in a number of samples; creating an enlarged minority class by generating new samples from the samples within the minority class and adding the new samples to the minority class; selecting subset samples from both the samples within the enlarged minority class and the majority class; weighting each of the subset samples based upon user input defining goals for attributes of a training sample set to be used in training the machine-learning model; and generating, using the neural network, the training sample set by re-running the selecting in view of the weighting.

Training sample set generation from imbalanced data in view of user goals

Aspects of the present disclosure relate to neural-based ontology generation and refinement. A set of input data can be received. A set of entities can be extracted from the set of input data using a named-entity recognition (NER) process, each entity having a corresponding label, the corresponding labels making up a label set. The label set can be compared to concepts in a set of reference ontologies. Labels that match to concepts in the set of reference ontologies can be selected as a candidate concept set. Relations associated with the candidate concepts within the set of reference ontologies can be identified as a candidate relation set. An ontology can then be generated using the candidate concept set and candidate relation set.

Neural-based ontology generation and refinement

Knowledge gaps in a chatbot are identified with reference to a domain-specific document and a set of QA pairs of the chatbot. Entities and/or entity values associated with the document are compared to the entities and/or entity values of the QA pairs. Entities of the document not associated with the QA pairs are identified as knowledge gaps. The QA pairs and knowledge gaps are ranked by relevance to the domain.

Growing community of inventors

Bengaluru, India

Hima Patel