Text and Multimedia Mining, 2020
Instructors:
- Martha Larson (linguistics, speech recognition, multimedia computing, recommendation and retrieval)
- Iris Hendrickx (NLP).
Schedule and deadlines
TMM: workload of about 10 hours per week with a high-pace schedule.
Part I:
- Lectures: 26h
- Reading/videos: 24h
- 4 Assignments: 24h
- 1 exam: 10h
Part II:
- Friday 30 October: Initial research idea
- Friday 27 November: Project plan
- Friday 29 January: Final deadline paper
Take-home exam:
Wed 9 December 23:59 - Sun 13 December 23:59
Materials
- ChengXiang Zhai & Sean Massung, "Text Data Management and Analysis (Official MOOC version), A Practical Introduction to Information Retrieval and Text Mining, e-book"
- Eisenstein, "Natural Language Processing", draft. 2018
- ... one more.
Exam material is everything including all assignments.
Assignments
Pass/fail; assignments may be discussed with others, but handed in work has to be your own.
10C system for self-check on your answer quality.
Content
W1: Introduction:
- Isayev - Text mining facilitates materials discovery
- TMM-1-Introduction.md
-
DIY Manual Text Mining Source
- DIY Automatic Text Mining Source
- DIY Characteristics of textual data Source
W2: Data preparation
- Eisenstein-Natural_Language_Processing.md
- Section 4.2, 4.3, Ch. 8.1-4
- TMM-2-NLP_Pipeline.md
- Zhai & Massung 2016: 3.3
W3: Word embeddings:
TODO:
- Zhai 2016
- 2.3 machine learning
- 14.3.3.2 word embeddings
- Ch. 15
W4: Information Extraction
- Eisenstein Section 17.1-17.4
- TMM-4-Information_extraction.md
- TMM-4a-Semantic_resources.md
W5: Multimedia
- TMM-5-Multimedia.md
- Interactive part; probably delete this note
- TMM-5a-Clustering.md
- TMM-5b-Semantic_Gap.md
- TMM-5c-Multimedia_Representations.md Source
- TMM-5d-AI_vs_Our_Eye.md Source
W6: Authorship Attribution
- TMM-6-Authorship_Attribution.md
- TMM-6b-Topic_modeling.md
- Recommended: Solorio 2013 on sockpuppet detection in Wikipedia.
TODO:
- DIY "Hildegard of Bingen: Authorship and Stylometry" by Mike Kestemont Source
W7: Sentiment Analysis
TODO:
- Ch.1 and 2.1 from Bing Liu - Sentiment Analysis and Opinion Mining, 2012.
W8a: Automatic Summarization
Recorded guest lectures, annoyingly in the exam period.
Automatic Summarization:
- Maya Sepelli: TMM-8a-Automatic_Summarization.md
- Yue Dong - Survey on Neural Network-Based Summarization Methods
- Above lecture discusses this paper
W8b: Topic Modelling
Topic Modelling (both done but no notes made):
- DIY on topic modelling Source
- Zhai Ch. 17
TODO:
- David M. Blei: Probabilistic Topic Models
Recommended: LDA in Python
W8c: Event detection
Event detection:
- TMM-8c-Event_detection.md
- Weiler A et al, 2016: Survey and experimental analysis of event detection techniques on twitter