Web and Social Information Extraction - A.Y. 2019/2020
The course presents algorithms and architectures to retrieve information from the web and to analyze social networks. Topics are information retrieval, web search engines, web mining, social network analytics.
IMPORTANT INFORMATION: From Monday, March 9th until the end of COVID-19 emergency, lessons will be REGULARLY held in streaming following the same schedule of in-class lessons. Students will receive the link to the videoconference few minutes before the beginning of the lesson. Information will be distributed through the course Google group and you will be able to participate in the videoconference ONLY using your institutional account. Videorecorded lessons and labs are available on a drive folder ONLY to students of the Google group.
Course schedule
II semester:
Monday |
14.00-16.00 |
aula alfa via Salaria 113 (Lab will be held in Colossus labs- via salaria 113 - Since 2nd week of March) |
Wednesday |
10.00-13.00 |
aula alfa via Salaria 113 |
Course Organization
The course presents architectures and algorithms related with the extraction of information from the Web, analyzing both Web Search engines and on-line Social Networks. A number of lessons are held in the Colossus lab.
During the lab, students will learn:
- Lucene, an open-source text-search library. With Lucene, you will learn to index and search a document archive
- How to useTwitter API, to track and index Twitter messages, to create and analyze word time series, and more
- To build a web scraper and trace the content of a forum
Self Assessment and Final Project
Self assessments and final project are sent to all students by email (google group).
Summary of Course Topics
Architecture of Information Retrieval systems. Tokenisation, stop-word removal and stemming; morphology; selection of index terms, use of thesauri. Inverted indices. Boolean and vector-space retrieval models, ranked retrieval and text-similarity metrics. Performance metrics: recall, precision, and F-measure. Evaluations on benchmark text collections. Latent Semantic Indexing. Relevance feedback. Query expansion.
Web Information Retrieval. Browsing and Scraping. Link analysis: Page Rank and HITS.
Social Network analysis: Opinion Mining, Social Network analysis, Community Detection, Social Media Analytics, Recommender systems
Textbooks
Exam
Standard grading policy:
- a) Written or oral exam on course material (50%). For those who attend classes, written test is split into a mid-term and a final test (usually, April and June).
- b) Project (teams of 2) (50%). The quality of developed software is matter of evaluation.
IMPORTANT: Due to COVID-19 emergency, the mid-term test will be replaced by a new grading policy, which includes an "interactivity grade" to prize active students (those who submit self assessments and lab exercises, and present their work during classes in streming).
The interactivity grade will be 1/3 of your final grade, and incorporates also interactivity and contributions during Labs.
Final grade will be computed as follows:
- 1/3 interactivity
- 1/3 project
- 1/3 final written test (hopefully in presence..)
Those with zero or low interactivity will be given the opportunity to recover the missing 1/3 (with a more complex written test ).
Project
The project is presented after the first 4-5 weeks of the course.
Find here, as an example, the best project in 2018:
wsie_report_moschella_spini.zip:
Project 2019
Find here the dataset for the project:
datasetwebsocial2019.zip
The description of the project can be downloaded from this link: <a data-saferedirecturl="https://www.google.com/url?q=https://docs.google.com/document/d/1NYVXM_X7qEX95RdFHnw0M3H2b-2BU_72IwmMxWVNca8/edit?usp%3Dsharing&source=gmail&ust=1559035338277000&usg=AFQjCNF9cfAdDVVAucDhoRLQa0NDtctAXw" href="https://docs.google.com/document/d/1NYVXM_X7qEX95RdFHnw0M3H2b-2BU_72IwmMxWVNca8/edit?usp=sharing" target="_blank">https://docs.google.com/document/d/1NYVXM_X7qEX95RdFHnw0M3H2b-2BU_72IwmMxWVNca8/edit?usp=sharing</a>.
best project in 2019 :
MiningEvolvingTopics_best_2019.pdf (code can be requested to Bardh Prenkaj)
Project 2018
The project is on Recommender Systems.
The description is provided in
Project_Description_2018.pdf
The description includes the link from where you can download the main dataset for the project, the Wiki_MID dataset.
The three additional datasets are:
S21.tsv S22_preferences.tsv and
S23.tsv
Project 2020
assigned (recommender Systems in Tourism)
Google Group
Slides and course materials (note: UPDATED = 2020)
Last Update | Topic | PPT | ZIP/PDF | Details | Suggested readings |
2020 | Introduction |
1.WS1.introAB.pptx (part 2)
1.WS1.introB.pptx (part 3) |
1.WS1.introA.pptx.zip (part 1)
| Introduction, architecture of IR systems, text processing, indexing |
A survey on Indexing Techniques for Big Data: Taxonomy and Performance Evaluation (2015) https://www.researchgate.net/publication/273082158_A_survey_on_Indexing_Techniques_for_Big_Data_Taxonomy_and_Performance_Evaluation
The case for learned index structures (April 30th, 2018) https://arxiv.org/pdf/1712.01208.pdf |
2020 | Basic Ranking Models | pptx | pdf | Basic ranking models: Boolean, Vector Space | https://nlp.stanford.edu/IR-book/pdf/01bool.pdf |
2020 | Query Expansion | ppt | | Improving basic IR models |
Query_Expansion_Techniques_for_Information_Retriev_1.pdf
QueryExpansionTechniquesSurvey.pdf |
2020 |
Eigenvectors, eigenvalues and SVD decomposition
Retrieval with Latent Semantic Indexing & Word Embeddings | 5b.Embeddings.pptx
|
5.LSI.pptx.zip
5.LSI.pdf.zip
5b.Embeddings.pdf
| Alternative Ranking Models based on word similarities |
https://cs224d.stanford.edu/lecture_notes/notes1.pdf
BERT: https://arxiv.org/pdf/1810.04805.pdf
(video lessons available on drive for Google Group) |
2020 | Evaluation | 3.EvaluationIR.pptx
| |
Performance measures and benchmarking of IR systems
|
https://papers.nips.cc/paper/5867-precision-recall-gain-curves-pr-analysis-done-right.pdf
see also: https://trec.nist.gov/ for TREC evaluation challenges
(video lessons available on drive for Google Group) |
2020 | Web Search | | 6.WebSearch-compressed.pdf | Web IR: crawling, scraping, searching on the Web; Categorization of web pages |
(video lessons available on drive for Google Group)
WEB_SITE_CLASSIFICATION_FEATURES_AND_ALG.pdf (survey on feature selection for web page classification)
information-10-00150.pdf (survey on machine learning for document classification ) |
2020 | Link Analysis | 7.link_analysis.pptx
| 7.link_analysis-compressed.pdf | Hyperlink-based Ranking (HITS and Page Rank) |
https://pdfs.semanticscholar.org/43b6/d922bcfcc8f8fcd3e7d22c8dc732653d9571.pdf
(video lessons available on drive for Google Group)
|
2020 | Social Media Analytics A1 | |
9a.SocialMediaAnalyticsA1.pdf
9a.SocialMediaAnalyticsA1.zip
| Social Network Analysis PART A1: node-centric measures of influence |
https://arxiv.org/pdf/1907.11229.pdf (a survey on trending topic detection from social networks ) https://www.researchgate.net/publication/277689549_Efficient_temporal_mining_of_micro-blog_texts_and_its_application_to_event_discovery (SAX*)
(video lessons available on drive for Google Group)
|
2020 | Social Media Analytics A2 | 9b.SocialMediaAnalyticsA2.pptx
| 9b.SocialMediaAnalyticsA2.pdf | Social Network Analysis PART A2: graph-based measures |
Surveytopknodesinsocialnetworks.pdf
(video lessons available on drive for Google Group)
|
2020 |
Social Media Analytics B
| 10.SocialMediaAnalytics_community_detectionB.pptx.zip
| 10.SocialMediaAnalytics_community_detectionB.pdf | Social Network Analysis PART B: community detection |
https://arxiv.org/ftp/arxiv/papers/1708/1708.00977.pdf
https://hal.archives-ouvertes.fr/file/index/docid/804234/filename/Survey-on-Social-Community-Detection-V2.pdf
(video lessons available on drive for Google Group)
|
2020 |
Social Media Analytics C
| 11.Maximizing-the-Spread-of-Influence.pptx
| 11.Maximizing-the-Spread-of-Influence.pdf | Spread of Influence in Social Networks |
Chapter 2 of https://wiki.eecs.yorku.ca/course_archive/2014-15/F/4412/_media/social_networks.pdf
(video lessons available on drive for Google Group)
|
2020 | Opinion Mining | | 13.OpinionMining.pdf | Searching for opinions on the Web | https://link.springer.com/content/pdf/10.1007/s10462-017-9599-6.pdf |
2020 | Recommender Systems | 12.RecommenderSystems.pptx
| 12.RecommenderSystems.pdf | Collaborative filtering, Content-based recommenders, Semantic recommenders |
http://shuaizhang.tech/2017/07/28/Summary-of-Recommender-System-Surveys-in-recent-years/ (a portal)
https://ieeexplore.ieee.org/document/8506344 ("A Survey of Collaborative Filtering-Based Recommender Systems: From Traditional Methods to Hybrid Methods Based on Social Networks" 2018)
(video lessons available on drive for Google Group) |
| LABs | NOTE: UDATED LAB MATERIAL, INCLUDING VIDEOs IS SHARED THROUGH THE GOOGLE GROUP | | Lucene, Crawlers, Scrapers, Social Media, Temporal series | (video labs available on drive for Google Group members) |
|
| Lab: Graph-G Library | | | Graph Libraries | |
Syllabus
Part A: Web Information Retrieval
- Introduction, Architecture of IR systems
- Text processing, Indexing
- Boolean and Vector Space Models
- Query expansion, understanding users' needs
- Ranking and query expansion based on word similarities (Singular value Decomposition, Word Embeddings)
- Evaluation methods: experimental and theoretical methods
- Web Information Retrieval
- Link Analysis
Part B:Social Information Extraction
- Social Network Analysis: graph-based measures, community detection, topic diffusion, temporal analysis
- Opinion Mining
- Recommender Systems
Topics for Final Dissertation
e-health, network medicine, event detection in social media and on the web, enterprise social networks, applications to social studies, temporal information retrieval, semantic recommenders, prediction and edetection of trendy topics
Merit-based (= high grades, very good programming skills) scholarships are available - ask the instructor-