Natural Language Processing with Pynorama
•Download as PPTX, PDF•
1 like•545 views
(Slides from my talk at the PyData 2018 conference) Visualising Natural Language Processing pipelines can be tricky, especially at scale. In this talk I will introduce Pynorama – a visualisation system that is written in Python, easy to setup, scalable, extensible, highly interactive and allows you to browse, analyse and understand your datasets and machine learning models.
Recommended
Recommended
More Related Content
Similar to Natural Language Processing with Pynorama
Similar to Natural Language Processing with Pynorama (20)
Recently uploaded
Recently uploaded (20)
Natural Language Processing with Pynorama
- 1. April 2018 Natural Language Processing Visualisation with Pynorama Slavi Marinov, Head of Machine Learning Technology – Man AHL 0
- 2. ©Man2018 Plan for today 1 • NLP primer • Pynorama demo • How to plug in your own datasets
- 3. ©Man2018 A typical text dataset 2Source: https://www.wsj.com/articles/developer-of-app-that-harvested-facebook-data-says-it-didnt-prove-useful-1524591027?mod=yahoo_hs&yptr=yahoo
- 4. ©Man2018 Working with text data 3 • Text dataset = Documents + Metadata • Can be quite large • Metadata is typically wide (100+ columns) • Many metrics are generated (even wider!) • NLP models are typically pipelines: • Raw document => Parsing => Tokenisation => Lemmatisation => Vectorisation => Classification
- 5. ©Man2018 Typical understanding tasks 4 • Exploration: Browse through the metadata and identify documents of interest • Exploration: Review a (random) sample of documents • Debugging: Trace a document through my pipeline • Introspection: Review extreme samples (e.g. sentiment) • Introspection: Review documents in certain cells of the confusion matrix (e.g. classification)
- 6. ©Man2018 The text visualisation ecosystem 5 • displacy: https://explosion.ai/demos/displacy • pyLDAvis: https://github.com/bmabey/pyLDAvis • facets: https://pair-code.github.io/facets/ • Embedding projector: https://projector.tensorflow.org/ • Prodigy: https://prodi.gy/docs/ • … and so many more • seriously: look at http://textvis.lnu.se/
- 7. ©Man2018 What I need 6 • Interactive UI to visualise my documents, metadata, and NLP pipelines • Fast, minimalistic, and easy to plug in • Extensible and scalable • Python* An intern project we can open-source! (Alex Wettig a.k.a. @CodeCreator)
- 9. ©Man2018 Pynorama core concepts [1/2] 8 Transforms Table Pipeline Source: Man Group
- 10. ©Man2018 Pynorama core concepts [2/2] 9 Pipeline Viewers Source: Man Group
- 11. ©Man2018 How to plug in? 10 • Implement 3 functions per dataset • get_table(): The table is a container for the metadata of the documents in the corpus (no text!) • get_pipeline(): The processing graph • get_record(): A document in a given stage of the pipeline
- 12. ©Man2018 11 • Return a Table instance • Tables have a length and know how to transform themselves • Multiple backends: Pandas (of course!), MongoDB (big datasets, lazy loading) • Out-of-the-box transformations: sample, sort, search, range query, remove nans, compute histogram get_table()
- 14. ©Man2018 13 • Return a dict representation of the graph • dict of nodes • key is node name • value contains the node type (associated with frontend Viewer) and the list of parents get_pipeline()
- 16. ©Man2018 15 • Tie table and pipeline together • Receives document key and pipeline node name • Return data for the frontend Viewer • Off the shelf Viewers: raw, pdf, xml, json, tree get_record(key, stage)
- 18. ©Man2018 Tying it all together 17Source: Man Group
- 19. ©Man2018 For extra credit 18 • load(): One-off initialisation of the dataset • get_config(): Frontend configuration
- 20. ©Man2018 19 • Initialise the data once for the duration of the server’s life • Internally, we cache in Arctic: https://github.com/manahl/arctic • User can reload on-demand for dynamic datasets load()
- 22. ©Man2018 21 • Encapsulate list of frontend configs • List of available transformations • List of initially visible columns • Name of column containing key (for get_record) • make_config() helper to… make it get_config()
- 24. ©Man2018 23 • Backend: Flask • Frontend: React + Redux • We didn’t even talk about the frontend today (more code there than the backend!) • Extensibility: transforms, viewers, tables, datasets Pynorama is modular
- 25. ©Man2018 24 • More tests • More advanced frontend Viewers • More advanced transformers • Nice caching story • Reload scheduling • Connect with other open-source NLP visualisers, e.g. displacy, pyLDAvis Pynorama is open source ;) Contributions welcome!
- 27. ©Man2018 26 • Transparent to the user • functions that take a Table and a metadata dict • e.g. for a sorting transform, the metadata may be the column name and the direction • Returns a new instance of the Table with the transform applied Appendix: A note on transforms
- 28. ©Man2018 Transform example 27Source: Man Group
- 29. ©Man2018 28 • Transparent to the user • Persist the state of the frontend • Which columns are visible • Which transforms are applied • Which viewers are open • Multiple backends: JSON on disk, Mongo Appendix: A note on sessions
- 30. ©Man2018 Important Information This information is communicated and/or distributed by the relevant Man entity identified below (collectively the “Company”) subject to the following conditions and restriction in their respective jurisdictions. Opinions expressed are those of the author and may not be shared by all personnel of Man Group plc (‘Man’). These opinions are subject to change without notice, are for information purposes only and do not constitute an offer or invitation to make an investment in any financial instrument or in any product to which the Company and/or its affiliates provides investment advisory or any other financial services. Any organisations, financial instrument or products described in this material are mentioned for reference purposes only which should not be considered a recommendation for their purchase or sale. Neither the Company nor the authors shall be liable to any person for any action taken on the basis of the information provided. Some statements contained in this material concerning goals, strategies, outlook or other non-historical matters may be forward-looking statements and are based on current indicators and expectations. These forward-looking statements speak only as of the date on which they are made, and the Company undertakes no obligation to update or revise any forward-looking statements. These forward-looking statements are subject to risks and uncertainties that may cause actual results to differ materially from those contained in the statements. The Company and/or its affiliates may or may not have a position in any financial instrument mentioned and may or may not be actively trading in any such securities. This material is proprietary information of the Company and its affiliates and may not be reproduced or otherwise disseminated in whole or in part without prior written consent from the Company. The Company believes the content to be accurate. However accuracy is not warranted or guaranteed. The Company does not assume any liability in the case of incorrectly reported or incomplete information. Unless stated otherwise all information is provided by the Company. Past performance is not indicative of future results. Australia: To the extent this material is distributed in Australia it is communicated by Man Investments Australia Limited ABN 47 002 747 480 AFSL 240581, which is regulated by the Australian Securities & Investments Commission (ASIC). This information has been prepared without taking into account anyone’s objectives, financial situation or needs. European Economic Area: Unless indicated otherwise this material is communicated in the European Economic Area by Man Solutions Limited which is an investment company as defined in section 833 of the Companies Act 2006 and is authorised and regulated by the UK Financial Conduct Authority (the “FCA”). Man Solutions Limited is registered in England and Wales under number 3385362 and has its registered office at Riverbank House, 2 Swan Lane, London, EC4R 3AD, England. As an entity which is regulated by the FCA, Man Solutions Limited is subject to regulatory requirements, which can be found at http://register.fca.org.uk. Germany: To the extent this material is used in Germany, the communicating entity is Man (Europe) AG, which is authorised and regulated by the Liechtenstein Financial Market Authority (FMA). Man (Europe) AG is registered in the Principality of Liechtenstein no. FL-0002.420.371-2. Man (Europe) AG is an associated participant in the investor compensation scheme, which is operated by the Deposit Guarantee and Investor Compensation Foundation PCC (FL-0002.039.614-1) and corresponds with EU law. Further information is available on the Foundation's website under HYPERLINK "http://www.eas-liechtenstein.li"www.eas-liechtenstein.li. This material is of a promotional nature. Hong Kong: To the extent this material is distributed in Hong Kong, this material is communicated by Man Investments (Hong Kong) Limited and has not been reviewed by the Securities and Futures Commission in Hong Kong. This material can only be communicated to intermediaries, and professional clients who are within one of the professional investor exemptions contained in the Securities and Futures Ordinance and must not be relied upon by any other person(s). 29
- 31. ©Man2018 Important Information Liechtenstein: To the extent the material is used in Liechtenstein, the communicating entity is Man (Europe) AG, which is regulated by the Financial Market Authority Liechtenstein (FMA). Man (Europe) AG is registered in the Principality of Liechtenstein no. FL-0002.420.371-2. Man (Europe) AG is an associated participant in the investor compensation scheme, which is operated by the Deposit Guarantee and Investor Compensation Foundation PCC (FL-0002.039.614-1) and corresponds with EU law. Further information is available on the Foundation's website under HYPERLINK "http://www.eas-liechtenstein.li"www.eas-liechtenstein.li. Switzerland: To the extent this material is distributed in Switzerland, this material is communicated by Man Investments AG, which is regulated by the Swiss Financial Market Authority FINMA. United States: To the extent his material is distributed in the United States, it is communicated and distributed by Man Investments, Inc. (‘Man Investments’). Man Investments is registered as a broker-dealer with the SEC and is a member of the Financial Industry Regulatory Authority (‘FINRA’). Man Investments is also a member of the Securities Investor Protection Corporation (‘SIPC’). Man Investments is a wholly owned subsidiary of Man Group plc. The registration and memberships described above in no way imply a certain level of skill or expertise or that the SEC, FINRA or the SIPC have endorsed Man Investments. Man Investments, 452 Fifth Avenue, 27th fl., New York, NY 10018. This material is proprietary information and may not be reproduced or otherwise disseminated in whole or in part without prior written consent. Any data services and information available from public sources used in the creation of this material are believed to be reliable. However accuracy is not warranted or guaranteed. 180717/RW/GL/R/W 30 continued
Editor's Notes
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL
- Technology is 1 of the 3 core drivers of all recent successes in ML It requires long-term investments in specialised hardware, software, data sets, and talent Has led to the rise of tightly integrated groups and hybrid researchers Creates its own risks to be understood and mitigated Examples from our work at AHL