Tata Consultancy Services Limited

Ashwin Srinivasan

Gautam Shroff

Puneet Kumar Agarwal

Lovekesh Vig

Sarmimala Saikia

Monika Sharma

Ehtesham Hassan

Rohit Rahul

Arindam Chowdhury

Gunjan Sehgal

Prerna Khurana

Mrinal Rawat

Vishwanath Doreswamy

Mohit Yadav

Mouli Rastogi

Syed Afshan Ali

Richa Rawat

Organizations are constantly flooded with questions, ranging from mundane to the unanswerable. It is therefore respective department that actively looks for automated assistance, especially to alleviate the burden of routine, but time-consuming tasks. The embodiments of the present disclosure provide BiLSTM-Siamese Network based Classifier for identifying target class of queries and providing responses to queries pertaining to the identified target class, which acts as an automated assistant that alleviates burden of answering queries in well-defined domains. Siamese Model (SM) is trained for a epochs, and then the same Base-Network is used to train Classification Model (CM) for b epochs iteratively until best accuracy is observed on validation test, wherein SM ensures it learns which sentences are similar/dissimilar semantically while CM learns to predict target class of every user query. Here a and b are assumed to be hyper parameters and are tuned for best performance on the validation set.

BiLSTM-siamese network based classifier for identifying target class of queries and providing responses thereof

This disclosure relates to a method and system for extracting information from images of one or more templatized documents. A knowledge graph with a fixed schema based on background knowledge is used to capture spatial and semantic relationships of entities present in scanned document and an adaptive lattice-based approach based on formal concepts analysis (FCA) is used to determine a similarity metric that utilizes both spatial and semantic information to determine if the structure of the scanned document image adheres to any of the known document templates. If a known document template whose structure is closely matching the structure of the scanned document is detected, then an inductive rule learning based approach is used to learn symbolic rules to extract information present in scanned document image and if a new document template is detected, then future scanned document images belonging to new document template are automatically processed using the learnt rules.

Methods and systems for extracting information from document images

Various methods are using SQL based data extraction for extracting relevant information from images. These are rule based methods of generating SQL-Query from NL, if any new English sentences are to be handled then manual intervention is required. Further becomes difficult for non-technical user. A system and method for extracting relevant from the images using a conversational interface and database querying have been provided. The system eliminates noisy effects, identifying the type of documents and detect various entities for diagrams. Further a schema is designed which allows an easy to understand abstraction of the entities detected by the deep vision models and the relationships between them. Relevant information and fields can then be extracted from the document by writing SQL queries on top of the relationship tables. A natural language based interface is added so that a non-technical user, specifying the queries in natural language, can fetch the information effortlessly.

Method and system for information extraction from document images using conversational interface and database querying

A method and system for interpreting a dataset is described herein. The method include computing a rule set pertaining to the dataset, followed by generating a rule cover pertinent to a subset of the rule set. Further, a plurality of distances between the plurality of rule pairs in the rule cover is calculated and a distance matrix based on the calculated plurality of distances is generated. Consequently, the overlapping rules within the rule cover are clustered using the distance matrix and a representative rule from each cluster is selected. Further, at least one exception for each representative rule is determined and the dataset is interpreted using the representative rules and the at least one exception. Thereby, the method provides succinct results in terms of rules and exceptions along with multiple interpretations of the same set of transactions from the dataset, thereby providing a holistic view about the dataset.

Interpretation of a dataset for co-occurring itemsets using a cover rule and clustering

This disclosure relates generally to multi-sensor data, and more particularly to summarizing multi-sensor data. In one embodiment, the method includes computing plurality of histograms from sensor data associated with a plurality of sensors. The respective histograms of each sensor are clustered into a first plurality of sensor-clusters, and a first set of rules is extracted therefrom. First set of rules defines patterns of histograms of a set of sensors occurring frequently over a time-period. Two or more sensor-clusters from amongst the first plurality of sensor-clusters are merged selectively to obtain a second plurality of sensor-clusters. Second set of rules are extracted from the second plurality of sensor-clusters, and a set of correlated sensors are identified therefrom based on the second set of rules. Third set of rules are extracted from the set of correlated sensors, the third set of rules summarizes the multi-sensor data to represent prominent co-occurring sensor behaviors.

Multi-sensor data summarization

Methods and systems for searching logical patterns in voluminous multi sensor data from the industrial internet is provided. The method retrieves instances of patterns in time-series data where patterns are specified logically, using a sequence of symbols. The logical symbols used are a subset of the qualitative abstractions specifically, the concepts of steady, increasing, decreasing. Patterns can include symbol-sequences for multiple sensors, approximate duration as well as slope values for each symbol. To facilitate efficient querying, each sensor time-series is pre-processed into a sequence of logical symbols. Each position in the resulting compressed sequence is registered across a TRIE-based index structure corresponding to the multiple logical patterns it may belong to. Logical multi-sensor patterns are efficiently retrieved and ranked using such a structure. This method of indexing and searching provides an efficient mechanism for exploratory analysis of voluminous multi-sensor data.

Methods and systems for searching logical patterns

Growing community of inventors

Sancoale, India

Ashwin Srinivasan