1 1 CSCI 5417 Information Retrieval Systems!Lecture 1 8/23/2011 Introduction 2What is Information Retrieval? Information retrieval is the science of searching for information in documents, searching for documents themselves, searching for metadata which describe documents, or searching within databases, whether relational stand-alone databases or hypertextually-networked databases such as the World Wide Web. Wikipedia The study of methods and structures used to represent and access information. Witten et al. The IR definition can be found in this book. Salton IR deals with the representation, storage, organization of, and access to information items. Salton Information retrieval is the term conventionally, though somewhat inaccurately, applied to the type of activity discussed in this volume. van Rijsbergen2 Manning et al. 3 Information Retrieval (IR) is finding material (usually documents) of an unstructured nature (usually text) that satisfies an information need from within large collections (usually stored on computers). 4How about: What you do when you use Google  Ad hoc retrieval is the core task that modern IR systems address  One-shot information seeking attempts by ignorant users  Ignorant about the structure and content of the collection  Ignorant about how the system works  Ignorant about how to formulate queries  Typically textual documents, but video and audio are becoming more prevalent  Collections are heterogeneous in nature3 But Web Search is Not All of IR  Specialist search (often boolean)  Research librarians  Medical retrieval  Legal search  Google scholar  MSN Academic search  Enterprise search  Social media  Twitter, facebook, etc  Desktop search  Apple’s Spotlight  Real time search  Twitter  Mobile search  Voice  Location aware search 56Social Media  Considerable interest right now lies in Web 2.0 issues...  Dealing with User-Generated Content  Discussion forums  Blogs  Microblogs  Social network sites  To deal with  Sentiment, opinions, etc  Social network structure  Location4 Web 78Course Plan  Cover the basics of IR technology in the first part of the course  The book provides the bulk of this material  Investigate newer topics in the latter part  Use discussions of real companies throughout the semester  Project presentations and discussions for the last section of the class.  I expect informed participation.5 9Course Plan  Core technology areas  Indexing and ranked retrieval  Basic vector space model  Probabilistic models  Supervised ML ranking methods  Document classification  Sometimes called routing or filtering  Supervised ML approaches  Document clustering  Unsupervised and semi-supervised ML approaches 10Administrivia  Work/Grading  Programming assignments 30%  Quizzes 30%  Project 30%  Participation 10%  Textbook  Introduction to Information Retrieval --- Manning, Raghavan and Schütze6 11Textbook  Lots of good stuff on the book’s website  Including the entire PDF of the book  I still recommend you buy it but it’s up to you  Based on my experience, people who buy it are more likely to read it  Last semester, people who had a physical copy got higher grades 12Administrivia  After the first assignments, the programming assignments will involve the use of Lucene (lucene.apache.org)  Open-source full text indexing system  Main Apache effort is Java  Various side efforts in Python, Ruby, C++, etc.  I don’t care which one you use  Your mileage may vary  Whether or not you use Lucene for the project is up to you7 13Lucene Documentation  Lucene has all the usual Java-doc style information associated with it. See the main Lucene page.  The main reference text associated with it is “Lucene in Action” 2ed.  See the publisher page for more info. Chapter 1 is free. 14CAETE  Remote students have a 1 week offset for assignments and quizzes  With warning I’m flexible on the assignments  Less so for the quizzes  For quizzes you need to have an EO or come to class for the quiz  The #1 problem that CAETE students run into is falling behind on the lectures  For what its worth, that’s also the #1 problem for local students who “attend” but are not prepared for class.8 15Web/Email  Slides will be available as ppt and pdf shortly after each class  You will be able to view the videos on the web. You should use this for review and for situations where you can’t come to class (like the flu), not as a reason to skip class  My roster has your colorado.edu email addresses. If you read your mail elsewhere then you need to set up a forward. Piazza  We’ll be using a new website as a resource for Q/A about the class.  Topics  Assignments  Quiz reviews etc.  Go to Piazza and register for this class  At least some part of “participation” can be taken care of through your use of Piazza 169 17Administrivia  Professor: Jim Martin  [email protected]  ECOT 726  Office hours TBA  www.cs.colorado.edu/~martin/csci5417/ 18Questions?10 19Simple Unstructured Data Scenario  Which plays of Shakespeare contain the words Brutus AND Caesar but NOT Calpurnia?  We could grep all of Shakespeare’s plays for Brutus and Caesar, then strip out lines containing Calpurnia. This is problematic:  Slow (for large corpora)  NOT Calpurnia is non-trivial  Other operations (e.g., find the word Romans near countrymen) not feasible  Lines vs Plays 20Grepping is Not an Option  So if we can’t search the documents in response to a query what can we do?  Create a data structure up front that will facilitate the kind of searching we want to do.11 21Term-Document Matrix 1 if play contains word, 0 otherwise Brutus AND Caesar but NOT Calpurnia 22Incidence Vectors  So we have a 0/1 vector for each term  Length of the term vector = number of plays  To answer our query: