F8S-2/CIS-2 Information Retrieval

Department of Computer Science and Engineering

F8S-2/CIS-2 Information Retrieval

Spring 2006

Daniel Ortiz-Arroyo

		Overview This is an introductory graduate level course on Information Retrieval. In this course we will study the main components, models and architecture of a modern Information Retrieval System. Information organization, text operations, and the metrics used to compare the performance of diverse systems will be described. We will also study the main models that have been proposed in Information Retrieval Systems to represent documents and queries and find the similarity between documents and queries. Finally, advanced techniques recently employed in IR will be presented. Exam and Questions: The 30 minutes exam will consists of 3 parts: 1) A 10 minute presentation of your assigned topic. Projector and a Laptop will be provided. Bring your presentation in a memory stick 2) Questions related to your topic 3) Questions that will be randomly selected from the 11 exercises posted in the table schedule shown below ( i.e. Exercise 1 up to Exercise 11). Time/Room/Schedule We'll start at 9AM each class. Our assigned classroom is B202. Please check the schedule regularly for new/updated information. The class notes posted below will be final the previous day before a class. The tentative schedule for the course is:
Week	Day	Class #	Subject	Exercises/Assignments	Class Notes*/Extra Reading Assignments
Week6	Tue	1	Introduction I: Historical perspective, information structure, information vs. data retrieval systems, digital libraries and IR systems organization. The WWW. Overview of the course.	Read Chapter 1 from Textbook Exercise 1 To be solved in class	Notes
Week6	Week6	Thur	2	Introduction II: Text operations: tokenization, stemming, stop words, lematization, compression. Words, terms and concepts, Thesaurus. Markup languages and the semantic web.	Read Chapter 7 and Appendix from Textbook Exercise 2 To be solved in class	Notes Read description of Porter's stemmer
Week7	Tue	3	Query types. Modeling in IR: Boolean and vector space models. Similarity measures.	Read Chapter 2.1-2.5.3 from Textbook Exercise 3 To be solved in class	Notes
Week7	Week7	Thur		CANCELLED	Look at Exercise 3 and try to solve it
Week 8	Tue	4	Review of probability concepts. Probabilistic model in IR.	Read Chapter 2.5.4, 2.6.2 from Textbook Exercise 4 To be solved in class	Notes Read this paper (at least up to Section 2.6)
Week8	Thur	5	Review of concepts in fuzzy logic. Fuzzy logic-based model in IR. Extended boolean model. Bayesian Networks in IR: the inference network model.	Read Chapter 2.6, 2.6.1 and 2.8.1-2.8.5 from Textbook Exercise 5 To be solved in class	Notes Read this paper
Week8	Week 9	Tue	6	Retrieval evaluation: Recall and Precision, alternate measures. Reference collections. Query Languages. Query Operations: pseudo-feedback local and global analysis.	Read Chapter 3,4,5 from Textbook Exercise 6 To be solved in class	Notes
Week 9	Thur		CANCELLED	Finish reading chapters 3,4,5 Look at exercise 6 and try to solve it
Week10	Tue	7	Ranking algorithms: HITS, PageRank. Indexing searching and storage mechanisms 1st part: flat, bitmap and signature files, PAT trees.	Read Sections 13.4.4 and 8.3 from Textbook Exercise 7	Notes Read this paper and this one
Week10	Week10	Thur	8	Indexing, searching and storage mechanisms 2nd part: Inverted files. Dictionaries: Tries and B-trees. Anatomy of search engines and crawlers. Libraries and Toolkits for IR: Lucene.	Read Sections 8.1, 8.2, and 13.1-13.4.3 from Textbook Exercise 8	Notes Read this article and this one
Week11	Tue 14th March	9	Efficient IR part I: Review of parallel processing. Flynn's classification. Speedup, efficiency, Amdhal's law and Amdhal's effect. Parallel and distributed mechanisms in search engines: data parallelism for logical and physical documents, data parallelism for terms.	Read Sections 9.1-9.2.2 and 9.3 from Textbook Exercise 9	Notes
Week11	Week11	Thur 16th March	10	Efficient IR part II: A parallel crawler. Static and dynamic partitioning of search graphs. Searching models. Multimedia IR: data modeling, queries and features. Searching and indexing multimedia objects using features: R-trees, GEMINI.	Read Chapter 11, sections 11.1-11.2.1,11.3.11.3.1. Chapter 12, sections 12.1-12.2 from Textbook Exercise 10	Notes Read this paper
Week12	Tue 21rd March		CANCELLED
Week12	Week12	Thur 23th March	11	Introduction to Artificial Intelligence and its application in IR systems. QA systems. Course overview. Final Exercise session	Exercise 11	Notes
Week13	Tue 28th March	12	Jens Rúni Poulsen Topic: Multiagent systems in IR Discussion	Presentation schedule and recommendations	Notes
Week13	Week13	Thur 30th March	13	CANCELLED
Week14	Tue 4th April	14	Ole Buus Topic: Advanced techniques in IR: genetic algorithms, simulated annealing etc. Daniel Jacob Poulsen Topic: Text classification(categorization) Discussion	Presentation schedule and recommendations	Notes on GA in IR systems by Ole Buus Notes on Text classification by D. J. Poulsen
Week14	Thur 6th April
Week15	Tue 11th April
	Thur 13th April
Week16	Tue 18th April
	Thur 20th April	15	Kim Beck Topic: Natural Language Processing in IR Jia Ma Topic: Machine Learning methods in IR: unsupervised/semi-supervised learning. Bing Pen Topic: Machine Learning methods in IR: supervised learning Discussion	Presentation schedule and recommendations	Notes on NLP in IR systems by Kim Beck Notes on unsupervised ML in IR by Jia Ma Notes on supervised ML in IR by Bing Pen
		Disclaimer: We will use slides/notes from the textbook, my own notes and from other sources on the web. References* •Textbook Modern Information Retrieval by Ricardo Baeza-Yates, Berthier Ribeiro-Neto Publisher: Addison Wesley; 1st edition (May 15, 1999) ISBN: 020139829X Reference books Mining the Web: Analysis of Hypertext and Semi Structured Data by Soumen Chakrabarti Publisher: Morgan Kaufmann; 1st edition (August 15, 2002) ISBN: 1558607544 Resources Porter's stemmer algorithm How Google works?