On Sentiment Analysis and Information Retrieval

Hi everyone,

we are very pleased to invite you to the third NLP Meetup where we will hear two interesting presentations.

=== Agenda ===

18:00 - Welcome reception
18:25 - [Introduction] Agenda and introduction
18:30 - [Presentation, incl. Q&A] Joint Aspect and Polarity Classification for Aspect-based Sentiment Analysis with End-to-End Neural Networks
19:15 - [Break] Networking and refreshment
19:45 - [Presentation, incl. Q&A] NLP and Information Retrieval
20:30 - [Networking] Get-together with sponsored food and drinks

=== FIRST Presentation ===

• Simon Steinheber: Maiborn Wolff

• Topic: Joint Aspect and Polarity Classification for Aspect-based Sentiment Analysis with End-to-End Neural Networks
In this talk, Simon will present a new model for aspect-based sentiment analysis that has been published under http://aclweb.org/anthology/D18-1139 .

In contrast to previous approaches, we jointly model the detection of aspects and the classification of their polarity in an end-to-end trainable neural network. We conduct experiments with different neural architectures and word representations on the recent GermEval 2017 dataset. We were able to show considerable performance gains by using the joint modeling approach in all settings compared to pipeline approaches. The combination of a convolutional neural network and fasttext embeddings outperformed the best submission of the shared task in 2017, establishing a new state of the art.

• Bio:
Simon studied computer science at LMU in Munich. His bachelor thesis (which concluded in the publication presented in this talk, http://aclweb.org/anthology/D18-1139) helped him to gain knowledge in the field of NLP and let him become an expert in Aspect-Based Sentiment-Analysis (ABSA). Today he is working at MaibornWolff on various ML/DL-Topics.

=== SECOND Presentation ===

• Christoph Goller: IntraFind

• Topic: NLP and Information Retrieval
In this talk, Christoph will present a general architecture for using NLP tools (such as text classification and named entity recognition, etc.) to improve information retrieval (full text search) based on Lucene / Elasticsearch / Solr. This includes a completely new approach for treating mixed language documents (based on language chunking), a linguistically enriched index containing Named Entities (including person names, physicochemical units, dates, sums of money, ...) and an extended Lucene query syntax for searching within such an enriched index. Most of these ideas are already implemented and available as a plugin for Elasticsearch. The next step they are currently working on is a component for analyzing natural language queries and translating them into queries for the enriched index in order to provide result lists or answers.

• Bio:
Christoph studied computer science at the Technical University of Munich where he also got his PhD in AI and Machine Learning. With his research he became one of the pioneers in Deep Learning. Christoph is an Apache Lucene Committer and has more than 20 years of experience in Information Retrieval, Natural Language Processing and AI.