[Pycon] [new paper] "Rocco Michele Lancellotti" - Social Network and External Communication Data analysis using NLP with industrial applications

[info a pycon.it]

Title: Social Network and External Communication Data analysis using NLP with industrial applications
Duration: 60 (includes Q&A)
Q&A Session: 0
Language: it
Type: Talk

Abstract: Audience Level
Data Scientists and Data Science Practitioners with basic to intermediate level experience in Topic Discovery, Named Entity Recognition and Social Network Analysis, basic Python programming skills.

Brief Description
Natural Language Processing is extensively used nowadays in industry. Applications can be found in customer service, reputation monitoring, advertising placement, market intelligence, regulatory compliance.
In this talk we show an industry application where data is extracted from social networks, online newspapers, forums, blogs and offline press all over the world. We implement Topic Discovery, Entity Recognition, Influencers Detection and Sentiment Analysis algorithms using Python, leading to a comprehensive model in terms of company’s reputation and facts. Results and insights are discussed from a business point of view. 

Abstract / Summary
Natural Language Processing is extensively used nowadays in industry to guide business executives through market difficulties, coming from customers, competitors and regulators actions and interactions. NLP business applications in industry can be found in the following areas:
•	customer service, using chatbots and online smart assistants which can provide immediate response to customer needs and decrease their pressure over human employees 
•	reputation monitoring, extracting meaningful informations from external and public data describing people opinions 
•	advertising placement, analyzing customer interests from written text 
•	market intelligence, creating structured databases of events about companies, governments and people 
•	regulatory compliance

Using Python, we show to the audience how NLP techniques can be applied to textual data extracted from social networks, online newspapers, forums, blogs and offline press. We use Topic Discovery, Entity Recognition, Influencers Detection and Sentiment Analysis algorithms implemented in Python, implementing both rule-based and statistical models, to extract precious informations in terms of brand reputation, market intelligence and customer service. We extend this work evaluating data written in different languages, highlighting pros and cons of chosen models applied to those languages. Finally, we discuss results/insights from a business point of view.


Tags: [u'open-data', u'nlp', u'sentiment-analysis', u'Social Network Analysis', u'industry-applications', u'entity-recognition', u'public-data', u'influencers-detection', u'topic-discovery']