2013.01.28 - SLIDE 1IS 240 – Spring 2013
Prof. Ray Larson University of California, Berkeley
School of Information
Principles of Information Retrieval
Lecture 2: Concepts and Elements
2013.01.28 - SLIDE 2IS 240 – Spring 2013
Review• Review
– IR History, Readings• Central Concepts in IR
– Documents– Queries– Collections– Evaluation– Relevance
• Elements of IR System
2013.01.28 - SLIDE 3IS 240 – Spring 2013
Review – IR History• Journal Indexes• “Information Explosion” following WWII
– Cranfield Studies of indexing languages and information retrieval
– Paper by Joyce and Needham on Thesauri for IR.
– Development of bibliographic databases • Chemical Abstracts • Index Medicus -- production and Medlars
searching
2013.01.28 - SLIDE 4IS 240 – Spring 2013
Development of IR Theory and Practice• Phase I: circa 1955-1975
– Foundational research– Fundamental IR concepts advanced in
research environment• Phase II: 1975 to present
– Slow adoption of IR research into operational systems
– Accelerated in mid-1990’s due to WWW search engines
2013.01.28 - SLIDE 5IS 240 – Spring 2013
Information Retrieval – Historical View
• Boolean model, statistics of language (1950’s)
• Vector space model, probablistic indexing, relevance feedback (1960’s)
• Probabilistic querying (1970’s)
• Fuzzy set/logic, evidential reasoning (1980’s)
• Regression, neural nets, inference networks, latent semantic indexing, TREC (1990’s)
• DIALOG, Lexus-Nexus, • STAIRS (Boolean based) • Information industry
(O($B))• Verity TOPIC (fuzzy logic)• Internet search engines
(O($100B?)) (vector space, probabilistic)
Research Industry
2013.01.28 - SLIDE 6IS 240 – Spring 2013
Readings and Discussion• Joyce and Needham
– Assigned index terms or Automatic?– Lattice theory (extension of Boolean algebra to
partially ordered sets)– Notice the Vector suggestion?
• Luhn– Document/Document similarity calculations based on
term frequency– KWIC indexes
• Doyle– Term associations
2013.01.28 - SLIDE 7IS 240 – Spring 2013
Readings (Next time)• Saracevic
– Relevance• Maron and Kuhns
– Probabilistic Indexing and matching• Cleverdon
– Evaluation• Salton and Lesk
– The SMART system• Hutchins
– Aboutness and indexing
2013.01.28 - SLIDE 8IS 240 – Spring 2013
Documents• What do we mean by a document?
– Full document?– Document surrogates?– Pages?
• Buckland (JASIS, Sept. 1997) “What is a Document”
• Bates (JASIST, June 2006) “Fundamental Forms of Information”
• Are IR systems better called Document Retrieval systems?
• A document is a representation of some aggregation of information, treated as a unit.
2013.01.28 - SLIDE 9IS 240 – Spring 2013
Collection• A collection is some physical or logical
aggregation of documents– A database– A Library– A index?– Others?
2013.01.28 - SLIDE 10IS 240 – Spring 2013
Queries• A query is some expression of a user’s
information needs• Can take many forms
– Natural language description of need– Formal query in a query language
• Queries may not be accurate expressions of the information need– Differences between conversation with a
person and formal query expression
2013.01.28 - SLIDE 11IS 240 – Spring 2013
User Information Need• Why build IR systems at all?• People have different and highly varied
needs for information• People often do not know what they want,
or may not be able to express it in a usable form– Filling the gaps in Boulding’s “Image”
• How to satisfy these user needs for information?
2013.01.28 - SLIDE 12IS 240 – Spring 2013
Controlled Vocabularies• Vocabulary control is the attempt to provide a
standardized and consistent set of terms (such as subject headings, names, classifications, or the thesauri discussed by Joyce and Needham) with the intent of aiding the searcher in finding information.
• Controlled vocabularies are a kind of metadata:– Data about data– Information about information
• Uncontrolled vs. Controlled– E.g., tagging images in Flickr
2013.01.28 - SLIDE 13IS 240 – Spring 2013
Pre- and Postcoordination• Precoordination relies on the indexer
(librarian, etc.) to construct some adequate representation of the meaning of a document.– E.g., “United States -- History -- Civil War,
1861-1865”• Postcoordination relies on the user or
searcher to combine more atomic concepts in the attempt to describe the documents that would be considered relevant.
2013.01.28 - SLIDE 14IS 240 – Spring 2013
Structure of an IR SystemSearchLine
Interest profiles& Queries
Documents & data
Rules of the game =Rules for subject indexing +
Thesaurus (which consists of
Lead-InVocabulary
andIndexing
Language
StorageLine
Potentially Relevant
Documents
Comparison/Matching
Store1: Profiles/Search requests
Store2: Documentrepresentations
Indexing (Descriptive and
Subject)
Formulating query in terms of
descriptors
Storage of profiles Storage of
Documents
Information Storage and Retrieval System
Adapted from Soergel, p. 19
2013.01.28 - SLIDE 15IS 240 – Spring 2013
Uses of Controlled Vocabularies• Library Subject Headings, Classification
and Authority Files.• Commercial Journal Indexing Services
and databases• Yahoo, and other Web classification
schemes• Online and Manual Systems within
organizations– SunSolve– MacArthur
2013.01.28 - SLIDE 16IS 240 – Spring 2013
Types of Indexing Languages• Uncontrolled Keyword Indexing• Folksonomies
– Uncontrolled but somewhat structured• Indexing Languages
– Controlled, but not structured• Thesauri
– Controlled and Structured• Classification Systems
– Controlled, Structured, and Coded• Faceted Classification Systems and Thesauri
2013.01.28 - SLIDE 17IS 240 – Spring 2013
Thesauri• A Thesaurus is a collection of selected
vocabulary (preferred terms or descriptors) with links among Synonymous, Equivalent, Broader, Narrower and other Related Terms
2013.01.28 - SLIDE 18IS 240 – Spring 2013
Development of a Thesaurus• Term Selection.• Merging and Development of Concept
Classes.• Definition of Broad Subject Fields and
Subfields.• Development of Classificatory structure• Review, Testing, Application, Revision.
2013.01.28 - SLIDE 19IS 240 – Spring 2013
Categorization Summary• Processes of categorization underlie many of
the issues having to do with information organization
• Categorization is messier than our computer systems would like
• Human categories have graded membership, consisting of family resemblances.
• Family resemblance is expressed in part by which subset of features are shared
• It is also determined by underlying understandings of the world that do not get represented in most systems
2013.01.28 - SLIDE 20IS 240 – Spring 2013
Classification Systems• A classification system is an indexing
language often based on a broad ordering of topical areas. Thesauri and classification systems both use this broad ordering and maintain a structure of broader, narrower, and related topics. Classification schemes commonly use a coded notation for representing a topic and it’s place in relation to other terms.
2013.01.28 - SLIDE 21IS 240 – Spring 2013
Classification Systems (cont.)• Examples:
– The Library of Congress Classification System– The Dewey Decimal Classification System– The ACM Computing Reviews Categories– The American Mathematical Society
Classification System
2013.01.28 - SLIDE 22IS 240 – Spring 2013
Evaluation• Why Evaluate?• What to Evaluate?• How to Evaluate?
2013.01.28 - SLIDE 23IS 240 – Spring 2013
Why Evaluate?• Determine if the system is desirable• Make comparative assessments• Others?
2013.01.28 - SLIDE 24IS 240 – Spring 2013
What to Evaluate?• How much of the information need is
satisfied.• How much was learned about a topic.• Incidental learning:
– How much was learned about the collection.– How much was learned about other topics.
• How inviting the system is.
2013.01.28 - SLIDE 25IS 240 – Spring 2013
What to Evaluate?What can be measured that reflects users’ ability to use system? (Cleverdon 66)– Coverage of Information– Form of Presentation– Effort required/Ease of Use– Time and Space Efficiency– Recall
• proportion of relevant material actually retrieved– Precision
• proportion of retrieved material actually relevant
effe
ctiv
enes
s
2013.01.28 - SLIDE 26IS 240 – Spring 2013
Relevance• In what ways can a document be relevant
to a query?– Answer precise question precisely.– Partially answer question.– Suggest a source for more information.– Give background information.– Remind the user of other knowledge.– Others ...
2013.01.28 - SLIDE 27IS 240 – Spring 2013
Relevance• “Intuitively, we understand quite well what
relevance means. It is a primitive “y’ know” concept, as is information for which we hardly need a definition. … if and when any productive contact [in communication] is desired, consciously or not, we involve and use this intuitive notion or relevance.”
» Saracevic, 1975 p. 324
2013.01.28 - SLIDE 28IS 240 – Spring 2013
Relevance• How relevant is the document
– for this user, for this information need.• Subjective, but• Measurable to some extent
– How often do people agree a document is relevant to a query?
• How well does it answer the question?– Complete answer? Partial? – Background Information?– Hints for further exploration?
2013.01.28 - SLIDE 29IS 240 – Spring 2013
Relevance Research and Thought
• Review to 1975 by Saracevic• Reconsideration of user-centered
relevance by Schamber, Eisenberg and Nilan, 1990
• Special Issue of JASIS on relevance (April 1994, 45(3))
2013.01.28 - SLIDE 30IS 240 – Spring 2013
Saracevic• Relevance is considered as a measure of
effectiveness of the contact between a source and a destination in a communications process– Systems view– Destinations view– Subject Literature view– Subject Knowledge view– Pertinence– Pragmatic view
2013.01.28 - SLIDE 31IS 240 – Spring 2013
Define your own relevance• Relevance is the (A) gage of relevance of
an (B) aspect of relevance existing between an (C) object judged and a (D) frame of reference as judged by an (E) assessor
• Where…
From Saracevic, 1975 and Schamber 1990
2013.01.28 - SLIDE 32IS 240 – Spring 2013
A. Gages• Measure • Degree• Extent• Judgement• Estimate• Appraisal• Relation
2013.01.28 - SLIDE 33IS 240 – Spring 2013
B. Aspect• Utility• Matching• Informativeness• Satisfaction• Appropriateness• Usefulness• Correspondence
2013.01.28 - SLIDE 34IS 240 – Spring 2013
C. Object judged• Document• Document representation• Reference• Textual form• Information provided• Fact• Article
2013.01.28 - SLIDE 35IS 240 – Spring 2013
D. Frame of reference• Question• Question representation• Research stage• Information need• Information used• Point of view• request
2013.01.28 - SLIDE 36IS 240 – Spring 2013
E. Assessor• Requester• Intermediary• Expert• User• Person• Judge• Information specialist
2013.01.28 - SLIDE 37IS 240 – Spring 2013
Schamber, Eisenberg and Nilan• “Relevance is the measure of retrieval
performance in all information systems, including full-text, multimedia, question-answering, database management and knowledge-based systems.”
• Systems-oriented relevance: Topicality• User-Oriented relevance• Relevance as a multi-dimensional concept
2013.01.28 - SLIDE 38IS 240 – Spring 2013
Schamber, et al. Conclusions• “Relevance is a multidimensional concept whose
meaning is largely dependent on users’ perceptions of information and their own information need situations
• Relevance is a dynamic concept that depends on users’ judgements of the quality of the relationship between information and information need at a certain point in time.
• Relevance is a complex but systematic and measureable concept if approached conceptually and operationally from the user’s perspective.”
2013.01.28 - SLIDE 39IS 240 – Spring 2013
Froelich• Centrality and inadequacy of Topicality as
the basis for relevance• Suggestions for a synthesis of views
2013.01.28 - SLIDE 40IS 240 – Spring 2013
Janes’ View
Topicality
Pertinence
Relevance
Utility
Satisfaction
2013.01.28 - SLIDE 41IS 240 – Spring 2013
Operational Definition of Relevance • From the point of view of IR evaluation (as
typified in TREC and other IR evaluation efforts)– Relevance is a term used for the relationship
between a users information need and the contents of a document where the user determines whether or not the contents are responsive to his or her information need
2013.01.28 - SLIDE 42IS 240 – Spring 2013
IR Systems• Elements of IR Systems• Overview – we will examine each of these
in further detail later in the course
2013.01.28 - SLIDE 43IS 240 – Spring 2013
What is Needed?• What software components are needed to
construct an IR system?
• One way to approach this question is to look at the information and data, and see what needs to be done to allow us to do IR
2013.01.28 - SLIDE 44IS 240 – Spring 2013
What, again, is the goal?• Goal of IR is to retrieve all and only the
“relevant” documents in a collection for a particular user with a particular need for information– Relevance is a central concept in IR theory
• OR• The goal is to search large document collections
(millions of documents) to retrieve small subsets relevant to the user’s information need
2013.01.28 - SLIDE 45IS 240 – Spring 2013
Collections of Documents…• Documents
– A document is a representation of some aggregation of information, treated as a unit.
• Collection– A collection is some physical or logical
aggregation of documents• Let’s take the simplest case, and say we
are dealing with a computer file of plain ASCII text, where each line represents the “UNIT” or document.
2013.01.28 - SLIDE 46IS 240 – Spring 2013
How to search that collection?• Manually?
– Cat, more• Scan for strings?
– Grep• Extract individual words to search???
– “tokenize” (a unix pipeline)• tr -sc ’A-Za-z’ ’\012’ < TEXTFILE | sort | uniq –c
– See “Unix for Poets” by Ken Church
• Put it in a DBMS and use pattern matching there…– assuming the lines are smaller than the text size limits
for the DBMS
2013.01.28 - SLIDE 47IS 240 – Spring 2013
What about VERY big files?• Scanning becomes a problem• The nature of the problem starts to change
as the scale of the collection increases• A variant of Parkinson’s Law that applies
to databases is:– Data expands to fill the space available to
store it • Currently this problem takes a new
approach – use MapReduce (like Hadoop)
2013.01.28 - SLIDE 48IS 240 – Spring 2013
The IR Approach• Extract the words (or tokens) along with
references to the record they come from– I.e. build an inverted file of words or tokens– Note that Google and others use MapReduce
approaches for this step• Is this enough?
2013.01.28 - SLIDE 49
Document Processing Steps
IS 240 – Spring 2013
2013.01.28 - SLIDE 50IS 240 – Spring 2013
What about …• The structure information, POS info, etc.?• Where and how to store this information?
– DBMS?– XML structured documents?– Special file structures
• DBMS File types (ISAM, VSAM, B-Tree, etc.)• PAT trees• Hashed files (Minimal, Perfect and Both)• Inverted files
• How to get it back out of the storage– And how to map to the original document location?
2013.01.28 - SLIDE 51IS 240 – Spring 2013
Structure of an IR SystemSearchLine Interest profiles
& QueriesDocuments
& data
Rules of the game =Rules for subject indexing +
Thesaurus (which consists of
Lead-InVocabulary
andIndexing
Language
StorageLine
Potentially Relevant
Documents
Comparison/Matching
Store1: Profiles/Search requests
Store2: Documentrepresentations
Indexing (Descriptive and
Subject)
Formulating query in terms of
descriptors
Storage of profiles Storage of
Documents
Information Storage and Retrieval System
Adapted from Soergel, p. 19
2013.01.28 - SLIDE 52IS 240 – Spring 2013
What next?• User queries
– How do we handle them?– What sort of interface do we need?– What processing steps once a query is
submitted?• Matching
– How (and what) do we match?
2013.01.28 - SLIDE 53IS 240 – Spring 2013
From Baeza-Yates: Modern IR…User Interface
Text operations
indexing DB Man.
Text Db
index
Queryoperations
Searching
Ranking
2013.01.28 - SLIDE 54IS 240 – Spring 2013
Query Processing• In order to correctly match queries and
documents they must go through the same text processing steps as the documents did when they were stored
• In effect, the query is treated like it was a document
• Exceptions (of course) include things like structured query languages that must be parsed to extract the search terms and requested operations from the query– The search terms must still go through the same text
process steps as the document…
2013.01.28 - SLIDE 55IS 240 – Spring 2013
Query Processing• Once the text is in a form to match to the
indexes then the fun begins– What approach to use?
• Boolean?• Extended Boolean?• Ranked
– Fuzzy sets?– Vector?– Probabilistic?– Language Models? – Neural nets?
• Most of the next few weeks will be looking at these different approaches
2013.01.28 - SLIDE 56IS 240 – Spring 2013
Display and formatting• Have to present the the results to the user• Lots of different options here, mostly
governed by – How the actual document is stored – And whether the full document or just the
metadata about it is presented