Oracle International Corporation

Oracle International Corportion

Vishal Vishnoi

Mark Edward Johnson

Thanh Long Duong

Srinivasa Phani Kumar Gadde

Elias Luqman Jalaluddin

Cong Duy Vu Hoang

Balakota Srinivas Vinnakota

Gautam Singaraju

Vladislav Blinov

Poorya Zaremoodi

Manish Parekh

Crystal C Pan

Yu-Heng Hong

Ying Xu

Xin Xu

Duy Vu

Tuyen Quang Pham

Xu Zhong

Aashna Devang Kanuga

Vishank Bhatia

Yakupitiyage Don Thanuja Samodhye Dharmasiri

Thanh Tien Vu

Saba Amsalu Teserra

Jae Min John

Gioacchino Tangari

Nitika Mathur

Philip Arthur

Suman Mallapura Somasundar

Boris Galitsky

Jiarui Ding

Muruganantham Chinnananchi

Shivashankar Subramanian

Stephen Andrew McRitchie

Mark Joseph Sugg

Steve Wai-Chun Siu

Edward E Wong

Anfernee Xu

Fen Wang

Stevan Malesevic

Sagar Gollamudi

Yuanxu Wu

King-Hwa Lee

Adam Pocock

Budhaditya Saha

Katherine Silverstein

Ahmed Ataallah Ataallah Abobakr

Dai Hoang Tran

Nagaraj N Bhat

Zikai Li

Xinwei Zhang

Ping L Lin

Alexander Wang

Syed Najam Abbas Zaidi

Christopher Kennewick

Chang Xu

Umanga Bista

Techniques are disclosed herein for improving model robustness on operators and triggering keywords in natural language to a meaning representation language system. The techniques include augmenting an original set of training data for a target robustness bucket by leveraging a combination of two training data generation techniques: (1) modification of existing training examples and (2) synthetic template-based example generation. The resulting set of augmented data examples from the two training data generation techniques are appended to the original set of training data to generate an augmented training data set and the augmented training data set is used to train a machine learning model to generate logical forms for utterances.

Model robustness on operators and triggering keywords in natural language to a meaning representation language system

Techniques are disclosed for training a model, using multi-task learning, to transform natural language to a logical form. In one particular aspect, a method includes accessing a first set of utterances that have non-follow-up utterances and a second set of utterances that have initial utterances and associated one or more follow-up utterances and training a model for translating an utterance to a logical form. The training is a joint training process that includes calculating a first loss for a first semantic parsing task based on one or more non-follow-up utterances from the first set of utterances, calculating a second loss for a second semantic parsing task based on one or more initial utterances and associated one or more follow-up utterances from the second set of utterances, combining the first and second losses to obtain a final loss, and updating model parameters of the model based on the final loss.

Transforming natural language to structured query language based on multi- task learning and joint training

Techniques are provided for generating training data to facilitate fine-tuning embedding models. Training data including anchor utterances is obtained. Positive utterances and negative utterances are generated from the anchor utterances. Tuples including the anchor utterances, the positive utterances, and the negative utterances are formed. Embeddings for the tuples are generated and a pre-trained embedding model is fine-tuned based on the embeddings. The fine-tuned model can be deployed to a system.

Training data generation to facilitate fine-tuning embedding models

A data corpus is partitioned into text strings for header classification. A group characteristic is computed for a text string, and whether the group characteristic satisfies a group characteristic criterion is determined. The text string may be disqualified from header classification if the group characteristic criterion is not satisfied, or one or more font characteristics may be determined for the text string if the group characteristic criterion is satisfied. A font characteristic that meets one or more prevalence criteria may be identified and evaluated to determine whether the font characteristic meets at least one font characteristic criterion. The text string may be disqualified from header classification if the font characteristic criterion is not satisfied, or if the font characteristic meets the font characteristic criterion, the text string is classified as a header, and tagged content is generated by applying a header tag to the text string.

Identifying and formatting headers for text content

Techniques for preprocessing data assets to be used in a natural language to logical form model based on scalable search and content-based schema linking. In one particular aspect, a method includes accessing an utterance, classifying named entities within the utterance into predefined classes, searching value lists within the database schema using tokens from the utterance to identify and output value matches including: (i) any value within the value lists that matches a token from the utterance and (ii) any attribute associated with a matching value, generating a data structure by organizing and storing: (i) each of the named entities and an assigned class for each of the named entities, (ii) each of the value matches and the token matching each of the value matches, and (iii) the utterance, in a predefined format for the data structure, and outputting the data structure.

Transforming natural language to structured query language based on scalable search and content-based schema linking

Techniques are disclosed for providing a scalable multi-tenant serve pool for chatbot systems. A query serving system (QSS) receives a request to serve a query for a new skillbot. The QSS comprises a plurality of deployments, each of which is configured to host a plurality of machine-learning models, each machine-learning model being associated with a skillbot, each deployment including a serving container and a model manager container that hosts a model manager, the serving container including a plurality of sub-containers, each of which hosts one of the machine-learning models downloaded by the model manager. The QSS selects a first deployment to be assigned to the new skillbot based on a first criterion, and loads the machine-learning model associated with the new skillbot into the first deployment. The machine-learning model is trained to serve the query for the new skillbot. The query is served using the machine-learning model.

Fast and scalable multi-tenant serve pool for chatbots

Techniques are disclosed for tuning hyperparameters of a model. Datasets are obtained for training the model and metrics are selected for evaluating performance of the model. Each metric is assigned a weight specifying an importance to the performance of the model. A function is created that measures performance based on the weighted metrics. Hyperparameters are tuned to optimize the model performance. Tuning the hyperparameters includes: (i) training the model that is configured based on a current values for the hyperparameters; (ii) evaluating a performance of the model using the function; (iii) determining whether the model is optimized for the metrics; (iv) in response to the model not being optimized, searching for a new values for the hyperparameters, reconfiguring the model with the new values, and repeating steps (i)-(iii) using the reconfigured model; and (v) in response to the model being optimized for the metrics, providing a trained model.

Method and system for constraint based hyperparameter tuning

Techniques are provided for using context tags in named-entity recognition (NER) models. In one particular aspect, a method is provided that includes receiving an utterance, generating embeddings for words of the utterance, generating a regular expression and gazetteer feature vector for the utterance, generating a context tag distribution feature vector for the utterance, concatenating or interpolating the embeddings with the regular expression and gazetteer feature vector and the context tag distribution feature vector to generate a set of feature vectors, generating an encoded form of the utterance based on the set of feature vectors, generating log-probabilities based on the encoded form of the utterance, and identifying one or more constraints for the utterance.

Context tag integration with named entity recognition models

Techniques are disclosed herein for synthesizing synthetic training data to facilitate training a natural language to logical form model. In one aspect, training data can be synthesized from original under a framework based on templates and a synchronous context-free grammar. In one aspect, training data can be synthesized under a framework based on a probabilistic context-free grammar and a translator. In one aspect, training data can be synthesized under a framework based on tree-to-string translation. In one aspect, the synthetic training data can be combined with original training data in order to train a machine learning model to translate an utterance to a logical form.

Data manufacturing frameworks for synthesizing synthetic training data to facilitate training a natural language to logical form model

The present disclosure relates to techniques for identifying out-of-domain utterances. One particular technique includes receiving an utterance and a target domain of a chatbot, generating a sentence embedding for the utterance, obtaining an embedding representation for each cluster of in-domain utterances associated with the target domain, predicting, using a metric learning model, a first probability that the utterance belongs to the target domain based on a similarity or difference between the sentence embedding and each embedding representation for each cluster, predicting, using an outlier detection model, a second probability that the utterance belongs to the target domain based on a determined distance or density deviation between the sentence embedding and embedding representations for neighboring clusters, evaluating the first probability and the second probability to determine a final probability, and classifying the utterance as in-domain or out-of-domain for the chatbot based on the final probability.

Techniques for out-of-domain (OOD) detection

A method includes receiving a training set of utterances for training a machine-learning model to identify one or more intents for one or more utterances, and augmenting the training set of utterances with out-of-domain (OOD) examples. The augmenting includes: generating a data set of OOD examples, filtering out OOD examples from the data set of OOD examples, determining a difficulty value for each OOD example remaining within the filtered data set of the OOD examples, and generating augmented batches of utterances including utterances from the training set of utterances and utterances from the filtered data set of the OOD based on the difficulty value for each OOD. Thereafter, the machine-learning model is trained using the augmented batches of utterances in accordance with a curriculum training protocol.

Out-of-domain data augmentation for natural language processing

Techniques are disclosed herein for focused training of language models and end-to-end hypertuning of the framework. In one aspect, a method is provided that includes obtaining a machine learning model pre-trained for language modeling, and post-training the machine learning model for various tasks to generate a focused machine learning model. The post-training includes: (i) training the machine learning model on an unlabeled set of training data pertaining to a task that the machine learning model was pre-trained for as part of the language modeling, and the unlabeled set of training data is obtained with respect to a target domain, a target task, or a target language, and (ii) training the machine learning model on a labeled set of training data that pertains to another task that is an auxiliary task related to a downstream task to be performed using the machine learning model or output from the machine learning model.

Framework for focused training of language models and techniques for end-to-end hypertuning of the framework

Techniques for maintaining list-type text formatting when converting content from a source content format to a destination content format are disclosed. A system generates text content by applying text formatting tags to segments of characters obtained from a source electronic document. The system parses a static-display type source electronic document to obtain character data of the characters in the source document. The system analyzes the parsed data to identify text arranged in a list-type text format in the source document. The system generates text content in a destination content format different from the source format by applying tags to segments of the text content designating the segments items in a list.

Generating tagged content from a list in an electronic document

Techniques are described for invoking and switching between chatbots of a chatbot system. In some embodiments, the chatbot system is capable of routing an utterance received while a user is already interacting with a first chatbot in the chatbot system. For instance, the chatbot system may identify a second chatbot based on determining that (i) such an utterance is an invalid input to the first chatbot or (ii) that the first chatbot is attempting to route the utterance to a destination associated with the first chatbot. Identifying the second chatbot can involve computing, using a predictive model, separate confidence scores for the first chatbot and the second chatbot, and then determining that a confidence score for the second chatbot satisfies one or more confidence score thresholds. The utterance is then routed to the second chatbot based on the identifying of the second chatbot.

Routing for chatbots

The present disclosure relates to chatbot systems, and more particularly, to batching techniques for handling unbalanced training data when training a model such that bias is removed from the trained machine learning model when performing inference. In an embodiment, a plurality of raw utterances is obtained. A bias eliminating distribution is determined and a subset of the plurality of raw utterances is batched according to the bias-reducing distribution. The resulting unbiased training data may be input into a prediction model for training the prediction model. The trained prediction model may be obtained and utilized to predict unbiased results from new inputs received by the trained prediction model.

Batching techniques for handling unbalanced training data for a chatbot

Growing community of inventors

Redwood City, CA, United States of America

Vishal Vishnoi