IP Library › Granted Patent US 12,423,620
Granted Patent B2
US 12,423,620 · App. 17/710,880 · Granted Sep 23, 2025

Intent-based task representation learning using weak supervision

Inventors: Oriana Riva (Redmond, WA); Michael Gamon (Seattle, WA); Sujay Kumar Jauhar (Kirkland, WA); Mei Yang (Redmond, WA); Sri Raghu Malireddi (Vancouver, CA); Timothy C. Franklin (Seattle, WA); Naoki Otani (Pittsburgh, PA)
Assignee: Microsoft Technology Licensing, LLC
G06N20/00G06F40/30
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,423,620
App. No.
17/710,880
Granted
Sep 23, 2025
Kind
B2
Abstract

Systems and methods are described that are generally directed to generating a general task embedding representing task information. In examples, the generated task embedding may include predicted task information such that, rather being underspecified, the task embedding representative of the task may include additional specified information, where the task embedding can then be utilized in many different models and applications. Thus, task data may be received and at least a portion of the task data may be encoded using an encoder. Based on one or more outputs generated by the encoder and a type embedding associated with the task data, a task intent may be extracted or otherwise predicted based on the task data and one or more type encodings associated with the task data. The intent extractor may be trained on multiple auxiliary tasks with weak supervision that provide semantic augmentation to under-specified task texts.

Claims (37)

1. A system, comprising:

a processor; and

memory, including machine-readable instructions, that when executed by the processor, cause the processor to:

receive task data representing a task;

encode at least a portion of the task data using an encoder;

generate a task embedding from an output generated by the encoder and a type embedding indicating a type of task associated with the task data, the task embedding including intent information indicating a predicted intent associated with the task data, wherein the task embedding is generated using a machine learning model trained on external semantically rich data sets that include contextual information describing tasks different than the task data representing the task, and the machine learning model is optimized over a plurality of auxiliary tasks; and

provide the generated task embedding to a receiving application, the generated task embedding including additional task-related data that is different than the task data representing the task.

2. The system of claim 1 , wherein the task data is underspecified and the encoder is configured to encode non-textual data about a portion of the underspecified task data.

3. The system of claim 2 , wherein the machine-readable instructions, when executed by the processor, cause the processor to combine one or more hidden states generated by the encoder with the type embedding indicating the type of task associated with the task data, wherein the one or more hidden states is the output generated by the encoder.

4. The system of claim 3 , wherein the non-textual data includes at least one of a due date, time, location of a user, importance level, and a person assigned to the underspecified task.

5. The system of claim 1 , wherein the intent information is obtained from an intent extractor and one or more parameters of the intent extractor are modified based on a predicted output obtained from at least one task of the auxiliary tasks.

6. The system of claim 1 , wherein the plurality of auxiliary tasks includes at least one of an autocompletion task, a pre-action and goal generation task, and an action arguments prediction task.

7. The system of claim 1 , wherein each auxiliary task of the plurality of auxiliary tasks generates an output based on the task embedding that is different than an output of another auxiliary task of the plurality of auxiliary tasks.

8. A method for generating a task embedding, the method comprising:

receiving task data representing a task;

encoding at least a portion of the task data using an encoder;

generating a task embedding from an output generated by the encoder and a type embedding indicating a type of task associated with the task data, the task embedding including intent information indicating a predicted intent associated with the task data, wherein the task embedding is generated using a machine learning model trained on external semantically rich data sets that include contextual information describing tasks different than the task data representing the task, and the machine learning model is optimized over a plurality of auxiliary tasks; and

provide the generated task embedding to a receiving application, the generated task embedding including additional task-related data that is different than the task data representing the task.

9. The method of claim 8 , wherein the task data is underspecified and the encoder is configured to encode non-textual data about a portion of the underspecified task data.

10. The method of claim 9 , further comprising:

combining one or more hidden states generated by the encoder with the type embedding associated with the task data, wherein the one or more hidden states is the output generated by the encoder.

11. The method of claim 10 , wherein the non-textual data includes at least one of a due date, time, location of a user, importance level, and a person assigned to the underspecified task.

12. The method of claim 8 , further comprising:

extracting the intent information using an intent extractor; and

modifying one or more parameters of the intent extractor based on a predicted output obtained from at least one task of the auxiliary tasks.

13. The method of claim 8 , wherein the plurality of auxiliary tasks includes at least one of an autocompletion task, a pre-action and goal generation task, and an action arguments prediction task.

14. The method of claim 8 , wherein each auxiliary task of the plurality of auxiliary tasks generates an output based on the task embedding that is different than an output of another auxiliary task of the plurality of auxiliary tasks.

15. A non-transitory computer-readable medium including instructions, that when executed by a processor, cause the processor to:

receive task data representing a to-do task;

encode at least a portion of the task data using an encoder;

generate a task embedding from an output generated by the encoder and a type embedding indicating a type of to-do task associated with the task data, the task embedding including intent information indicating a predicted intent associated with the task data, wherein the task embedding is generated using a machine learning model trained on external semantically rich data sets that include contextual information describing tasks different than the task data representing the to-do task, and the machine learning model is optimized over a plurality of auxiliary tasks; and

provide the generated task embedding to a receiving application, the generated task embedding including additional task-related data that is different than the task data representing the to-do task.

16. The computer-readable medium of claim 15 , wherein the task data is underspecified and the encoder is configured to encode non-textual data about a portion of the underspecified task data.

17. The computer-readable medium of claim 16 , wherein the instructions, when executed by the processor, cause the processor to combine one or more hidden states generated by the encoder with a type embedding of the one or more type embeddings associated with the task data, wherein the one or more hidden states is the output generated by the encoder.

18. The computer-readable medium of claim 17 , wherein the non-textual data includes at least one of a due date, time, location of a user, importance level, and a person assigned to the underspecified task.

19. The computer-readable medium of claim 17 , wherein the plurality of auxiliary tasks includes at least one of an autocompletion task, a pre-action and goal generation task, and an action arguments prediction task, and each auxiliary task of the plurality of auxiliary tasks generates an output based on the task embedding that is different than an output of another auxiliary task of the plurality of auxiliary tasks.

20. The computer-readable medium of claim 15 , wherein the intent information is obtained from an intent extractor and one or more parameters of the intent extractor are modified based on a predicted output obtained from at least one task of the auxiliary tasks.

Assignments (3)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Apr 12, 2022
From: FRANKLIN, TIMOTHY C.
To: MICROSOFT TECHNOLOGY LICENSING, LLC
Reel/Frame 059574/0513 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Apr 1, 2022
From: GAMON, MICHAEL; JAUHAR, SUJAY KUMAR; MALIREDDI, SRI RAGHU; OTANI, NAOKI; RIVA, ORIANA
To: MICROSOFT TECHNOLOGY LICENSING, LLC
Reel/Frame 059466/0448 →
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Apr 1, 2022
From: YANG, MEI
To: MICROSOFT TECHNOLOGY LICENSING, LLC
Reel/Frame 059475/0979 →
Continuity (2)
Provisional Application 63305191 · Jan 31, 2022
Related Publication 20230244989A1 · Aug 3, 2023
References Cited (63)
US 8694355B2 · Bui et al. · 2014 [cited by applicant]
US 10706373B2 · Gruber et al. · 2020 [cited by applicant]
US 20200125944A1 · Jauhar et al. · 2020 [cited by applicant]
US 20210142164A1 · Liu et al. · 2021 [cited by applicant]
CA 3123387A1 · 2021 [cited by applicant]
CN 113282368A · 2021 [cited by examiner]
WO WO2022023385A1 · 2022 [cited by examiner]
WO WO2022088444A1 · 2022 [cited by examiner]
Aghajanyan, et al., “Muppet: Massive Multi-task Representations with Pre-Finetuning”, In Repository of arXiv:2101.11038v1, Jan. 26, 2021, 12 Pages. [cited by applicant]
Barr, et al., “The Linguistic Structure of English Web-Search Queries”, In Proceedings of the Conference on Empirical Methods in Natural Language Processing, Oct. 25, 2008, pp. 1021-1030. [cited by applicant]
Bosselut, et al., “COMET: Commonsense Transformers for Automatic Knowledge Graph Construction”, In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, Jul. 28, 2019, pp. 4762-4779. [cited by applicant]
Broder, Andrei, “A Taxonomy of Web Search”, In Proceedings of ACM SIGIR Forum, vol. 36, Issue 2, Sep. 1, 2002, pp. 3-10. [cited by applicant]
Camacho-Collados, et al., “From Word To Sense Embeddings: A Survey on Vector Representations of Meaning”, In Journal of Artificial Intelligence Research, vol. 63, Dec. 6, 2018, pp. 743-788. [cited by applicant]
Chaudhari, et al., “An Attentive Survey of Attention Models”, In Journal of ACM Transactions on Intelligent Systems and Technology, vol. 12, Issue 5, Oct. 22, 2021, 32 Pages. [cited by applicant]
Cho, et al., “Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation”, In Proceedings of the Conference on Empirical Methods in Natural Language Processing, Oct. 25, 2014, pp. 1724… [cited by applicant]
Devlin, et al., “BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding”, In Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics: Human … [cited by applicant]
Dror, et al., “Replicability Analysis for Natural Language Processing: Testing Significance with Multiple Datasets”, In Journal of Transactions of the Association for Computational Linguistics, Dec. 1, 2017, pp. 471-486. [cited by applicant]
Du, et al., “General Purpose Text Embeddings from Pre-Trained Language Models for Scalable Inference”, In Repository of arXiv:2004.14287v1, Apr. 29, 2020, 12 Pages. [cited by applicant]
Gil, et al., “Capturing Common Knowledge about Tasks: Intelligent Assistance for To-Do Lists”, In Journal of ACM Transactions on Interactive Intelligent Systems, vol. 2, Issue 3, Sep. 1, 2012, 35 Pages. [cited by applicant]
Gorman, et al., “We need to talk about standard splits”, In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, Jul. 28, 2019, pp. 2786-2791. [cited by applicant]
Graus, et al., “Analyzing and Predicting Task Reminders”, In Proceedings of the Conference on User Modeling Adaptation and Personalization, Jul. 13, 2016, pp. 7-15. [cited by applicant]
Honnibal, et al., “spaCy: Industrial strength natural language processing in Python”, Retrieved from: https://github.com/explosion/spaCy/tree/v3.2.0, Nov. 5, 2021, 7 Pages. [cited by applicant]
Hwang, et al., “Comet-Atomic 2020: On Symbolic and Neural Commonsense Knowledge Graphs”, In Proceedings of the Thirty-Fifth AAAI Conference on Artificial Intelligence, Feb. 2, 2021, pp. 6384-6392. [cited by applicant]
Jauhar, et al., “Ms-latte: A dataset of where and when to-do tasks are completed”, In Repository of arXiv:2111.06902v1, Nov. 12, 2021, 10 Pages. [cited by applicant]
Keyaki, et al., “Part-of-speech Tagging for Web Search Queries Using a Large-scale Web Corpus”, In Proceedings of the Symposium on Applied Computing, Apr. 3, 2017, pp. 931-937. [cited by applicant]
Kobayashi, et al., “Attention is Not Only a Weight: Analyzing Transformers with Vector Norms”, In Proceedings of the Conference on Empirical Methods in Natural Language Processing, Nov. 16, 2020, pp. 7057-7075. [cited by applicant]
Landes, et al., “A Supervised Approach To The Interpretation Of Imperative To-Do Lists”, In Repository of arXiv:1806.07999v1, Jun. 20, 2018, 10 Pages. [cited by applicant]
Landes, Paul, “Supervised Approach Imperative To-Do List Categorization”, Retrieved from: https://github.com/plandes/todo-task, Jun. 29, 2018, 9 Pages. [cited by applicant]
Lewis, et al., “BART: Denoising Sequence-to-Sequence Pre-training for Natural Language Generation, Translation, and Comprehension”, In Proceedings of the 58th Annual Meeting of the Association for Computational Linguist… [cited by applicant]
Liu, et al., “Multi-Task Deep Neural Networks for Natural Language Understanding”, In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, Jul. 28, 2019, pp. 4487-4496. [cited by applicant]
Liu, et al., “Roberta: A Robustly Optimized BERT Pretraining Approach”, In Repository of arXiv:1907.11692v1, Jul. 26, 2019, 13 Pages. [cited by applicant]
Loshchilov, et al., “Decoupled Weight Decay Regularization”, In Journal of The Seventh International Conference on Learning Representations, May 6, 2019, 18 Pages. [cited by applicant]
Luong, et al., “Effective Approaches to Attention-based Neural Machine Translation”, In Proceedings of the Conference on Empirical Methods in Natural Language Processing, Sep. 17, 2015, pp. 1412-1421. [cited by applicant]
Maaten, et al., “Visualizing Data Using t-SNE”, In Journal of Machine Learning Research, vol. 9, Nov. 2008, pp. 2579-2605. [cited by applicant]
Mikolov, et al., “Distributed Representations of Words and Phrases and their Compositionality”, In Journal of Advances in Neural Information Processing Systems, vol. 26, Dec. 5, 2013, 9 Pages. [cited by applicant]
Mou, et al., “Natural Language Inference by Tree-Based Convolution and Heuristic Matching”, In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics, Aug. 7, 2016, pp. 130-136. [cited by applicant]
Mukherjee, et al., “Smart To-Do: Automatic Generation of To-Do Items from Emails”, In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, Jul. 5, 2020, pp. 8680-8689. [cited by applicant]
Nguyen, et al., “BERTweet: A pre-trained language model for English Tweets”, In Proceedings of the Conference on Empirical Methods in Natural Language Processing: System Demonstrations, Nov. 16, 2020, pp. 9-14. [cited by applicant]
Panchenko, et al., “Building a Web-Scale Dependency-Parsed Corpus from CommonCrawl”, In Proceedings of the Eleventh International Conference on Language Resources and Evaluation, May 7, 2018, pp. 1816-1823. [cited by applicant]
Pappas, et al., “GILE: A Generalized Input-Label Embedding for Text Classification”, In Transactions of the Association for Computational Linguistics, vol. 7, Apr. 2019, pp. 139-155. [cited by applicant]
Paszke, et al., “PyTorch: An Imperative Style, High-Performance Deep Learning Library”, In Proceedings of the 33rd Conference on Neural Information Processing Systems, Dec. 8, 2019, 12 Pages. [cited by applicant]
Patel, et al., “Magnitude: A Fast, Efficient Universal Vector Embedding Utility Package”, In Proceedings of the Conference on Empirical Methods in Natural Language Processing: System Demonstrations, Oct. 31, 2018, pp. 1… [cited by applicant]
Pedregosa, et al., “Scikit-learn: Machine Learning in Python”, In Journal of Machine Learning Research, vol. 12, Nov. 1, 2011, pp. 2825-2830. [cited by applicant]
Pentyala, et al., “Multi-Task Networks with Universe, Group, and Task Feature Learning”, In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, Jul. 28, 2019, pp. 820-830. [cited by applicant]
Pereyra, et al., “Regularizing neural networks by penalizing Confident output distributions”, In Journal of the Fifth International Conference on Learning Representations, Apr. 24, 2017, pp. 1-11. [cited by applicant]
Qiu, et al., “Pre-trained models for natural language processing: A survey”, In Journal of Science China Technological Sciences, vol. 63, Issue 10, Sep. 15, 2020, 25 Pages. [cited by applicant]
Nouri, et al., “Step-wise Recommendation for Complex Task Support”, In Proceedings of the Conference on Human Information Interaction and Retrieval, Mar. 14, 2020, pp. 203-212. [cited by applicant]
Radford, et al., “Language Models are Unsupervised Multitask Learners”, In Journal of OpenAI Blog, vol. 1, Issue 8, Feb. 24, 2019, 24 Pages. [cited by applicant]
Raffel, et al., “Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer”, In Journal of Machine Learning Research, vol. 21, Issue 140, Jun. 2020, pp. 1-67. [cited by applicant]
Ruder, Sebastian, “An Overview of Multi-Task Learning in Deep Neural Networks”, Retrieved from: https://ruder.io.multi-task/, May 29, 2017, 32 Pages. [cited by applicant]
Ruppenhofer, et al., “FrameNet II: Extended theory and practice”, Retrieved from: https://framenet2.icsi.berkeley.edu/docs/r1.7/book.pdf, Nov. 1, 2016, 129 Pages. [cited by applicant]
Shah, et al., “Bridging Task Expressions and Search Queries”, In Proceedings of the Conference on Human Information Interaction and Retrieval, Mar. 14, 2021, pp. 319-323. [cited by applicant]
Shui, et al., “A Principled Approach for Learning Task Similarity in Multitask Learning”, In Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence, Mar. 21, 2019, pp. 3446-3452. [cited by applicant]
Shuster, et al., “The Dialogue Dodecathlon: Open-Domain Knowledge and Image Grounded Conversational Agents”, In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, Jul. 5, 2020, pp. … [cited by applicant]
Speer, et al., “ConceptNet 5.5: An Open Multilingual Graph of General Knowledge”, In Proceedings of the AAAI Conference on Artificial Intelligence, Feb. 12, 2017, pp. 4444-4451. [cited by applicant]
Stickland, et al., “BERT and PALs: Projected Attention Layers for Efficient Adaptation in Multi-Task Learning”, In Proceedings of the 36th International Conference on Machine Learning, Jun. 9, 2019, 12 Pages. [cited by applicant]
Swayamdipta, et al., “Frame-Semantic Parsing with Softmax-Margin Segmental RNNs and a Syntactic Scaffold”, In Repository of arXiv:1706.09528v1, Jun. 29, 2017, 12 Pages. [cited by applicant]
Taghavi, et al., “An analysis of web proxy logs with query distribution pattern approach for search engines”, In Journal of Computer Standards & Interfaces, vol. 34, Issue 1, Jan. 2012, pp. 162-170. [cited by applicant]
Zhang, et al., “Multi-Task Learning for Sentiment Analysis with Hard-Sharing and Task Recognition Mechanisms”, In Journal of Information, vol. 12, No. 5, May 12, 2021, 13 Pages. [cited by applicant]
Vaswani, et al., “Attention Is All You Need”, In Proceedings of 31st Conference on Neural Information Processing Systems, Dec. 4, 2017, pp. 1-11. [cited by applicant]
Wolf, et al., “Transformers: State-of-the-Art Natural Language Processing”, In Proceedings of the Conference on Empirical Methods in Natural Language Processing: System Demonstrations, Nov. 16, 2020, pp. 38-45. [cited by applicant]
Zhang, et al., “Learning to Decompose and Organize Complex Tasks”, In Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Jun. 6, 20… [cited by applicant]
“International Search Report and Written Opinion Issued in PCT Application No. PCT/US22/049337”, Mailed Date: Mar. 27, 2023, 14 Pages. [cited by applicant]