1. Introduction to Digital Libraries Searching

2. Technical View: Retrieval as Matching Documents to Queries
Algorithm
Surrogates
Surrogates
Document
Space
Terms
Query Form A
Query
Space
Sample
Sample
Vectors
Query Form B
NO User!
No Feedback!
Etc..
Etc..
Retrieval is algorithmic. Evaluation is typically a binary
decision for each pairwise match and one or more aggregate values for a set of matches (e.g., recall and precision). 5

3. Human View: Information-Seeking Process
Results
Perceived
Needs
Queries
Indexes
Problem
Actions
Data
Physical
Interface
Information seeking is an active, iterative process controlled by a human who Changes throughout the process. Evaluation is relative to human needs. 6

4. IR Models s e Adhoc r Filtering T a k Browsing Extended Boolean vector
Set Theoretic
Fuzzy
Extended Boolean
Classic Models
boolean
vector
probabilistic
Algebraic
Generalized Vector
Lat. Semantic Index
Neural Networks U s e r T a k
Retrieval:
Adhoc
Filtering
Non-Overlapping Lists
Proximal Nodes
Structured Models
Probabilistic
Inference Network
Belief Network
Browsing
Browsing
Flat
Structure Guided
Hypertext

5. “Classic” Retrieval Models
Boolean
Documents and queries are sets of index terms
Vector
Documents and queries are documents in N-dimensional space
Probabilistic
Based on probability theory

6. Boolean Searching Exactly what you would expect
and, or, not operations defined
requires an exact match
based on inverted file
(computer and science) and (not(animals)) would prevent a document with “use of computers in animal science research” from being retrieved

7. Boolean ‘AND’
Information AND Retrieval
Information
Retrieval

8. Example Draw a Venn diagram for: What is the meaning of:
Care and feeding and (cats or dogs)
What is the meaning of:
Information and retrieval and performance or evaluation

9. Exercise D1 = “computer information retrieval”
D2 = “computer retrieval”
D3 = “information”
D4 = “computer information”
Q1 = “information  retrieval”
Q2 = “information  ¬computer”

10. Boolean-based Matching
Exact match systems; separate the documents containing a given term from those that do not.
Terms
Queries
Mediterranean
adventure
agriculture
cathedrals
horticulture
scholarships
disasters
bridge
leprosy
recipes
flags
tennis
Venus
flags AND tennis
leprosy AND tennis
Venus OR (tennis AND flags)
Documents
(bridge OR flags) AND tennis

11. Exercise 1
Swift 2
Shakespeare 3 4
Milton 5 6 7 8
Chaucer 9 10 11 12 13 14 15
((chaucer OR milton) AND (NOT swift)) OR ((NOT chaucer) AND (swift OR shakespeare))

12. Boolean features Order dependency of operators Nesting of search terms
( ), NOT, AND, OR (DIALOG)
May differ on different systems
Nesting of search terms
Nutrition and (fast or junk) and food

13. Boolean Limitations Searches can become complex for the average user
too much ANDing can clobber recall
tricky syntax:
“research AND NOT computer science”
“research AND NOT (computer science)” (implicit OR)
“research AND NOT (computer AND science)”
all different -- (frequently seen in NTRS logs)

14. Vector Model Calculate degree of similarity between document and query
Ranked output by sorting similarity values
Also called ‘vector space model’
Imagine your documents as N-dimensional vectors (where N=number of words)
The “closeness” of 2 documents can be expressed as the cosine of the angle between the two vectors

15. Vector Space Model
Documents and queries are points in N-dimensional space (where N is number of unique index terms in the data collection) Q D

16. Vector Space Model with Term Weights
assume document terms have different values for retrieval
therefore assign weights to each term in each document
example:
proportional to frequency of term in document
inversely proportional to frequency of term in collection

17. Graphic Representation
Example:
D1 = 2T1 + 3T2 + 5T3
D2 = 3T1 + 7T2 + T3
Q = 0T1 + 0T2 + 2T3 T3 T1 T2
D1 = 2T1+ 3T2 + 5T3
D2 = 3T1 + 7T2 + T3
Q = 0T1 + 0T2 + 2T3 7 3 2 5
Is D1 or D2 more similar to Q?
How to measure the degree of similarity? Distance? Angle? Projection?

18. Document and Query Vectors
Documents and Queries are vectors of terms
Vectors can use binary keyword weights or assume 0-1 weights (term frequencies)
Example terms: “dog”,”cat”,”house”, “sink”, “road”, “car”
Binary: (1,1,0,0,0,0), (0,0,1,1,0,0)
Weighted: (0.01,0.01, 0.002, 0.0,0.0,0.0)

19. Document Collection Representation
A collection of n documents can be represented in the vector space model by a term-document matrix.
An entry in the matrix corresponds to the “weight” of a term in the document; zero means the term has no significance in the document or it simply doesn’t exist in the document.
T1 T2 … Tt
D1 w11 w21 … wt1
D2 w12 w22 … wt2
: : : :
Dn w1n w2n … wtn

20. Inner Product: Example 1k1 k2 k3

21. factors information help human operation retrieval systems
Vector Space Example
Simple Match
Query ( )
Rec1 ( )
( ) =4
Rec2 ( )
( ) =3
Rec3 ( )
( ) =2
indexed words:
factors information help human operation retrieval systems
Query: human factors in information retrieval systems
Vector: ( )
Record 1 contains: human, factors, information, retrieval
Vector: ( )
Record 2 contains: human, factors, help, systems
Vector: ( )
Record 3 contains: factors, operation, systems
Vector: ( )
Weighted Match
Query ( )
Rec1 ( )
( ) =13
Rec2 ( )
( ) =8
Rec3 ( )
( ) =3

22. Term Weights: Term Frequency
More frequent terms in a document are more important, i.e. more indicative of the topic.
fij = frequency of term i in document j
May want to normalize term frequency (tf) across the entire corpus:
tfij = fij / max{fij}

23. Some formulas for Sim
Dot product
Cosine
Dice
Jaccard t1 D Q t2

24. Example
Documents: Austen's Sense and Sensibility, Pride and Prejudice; Bronte's Wuthering Heights
cos(SAS, PAP) = .996 x x x 0.0 = 0.999
cos(SAS, WH) = .996 x x x .254 = 0.929

25. Extended Boolean Model
Boolean model is simple and elegant.
But, no provision for a ranking
As with the fuzzy model, a ranking can be obtained by relaxing the condition on set membership
Extend the Boolean model with the notions of partial matching and term weighting
Combine characteristics of the Vector model with properties of Boolean algebra

26. The Idea qor = kx  ky; wxj = x and wyj = y ky (1,1) dj+1 OR dj
y = wyj
(0,0)
x = wxj kx
We want a document to be
as far as possible from (0,0)
sim(qor,dj) = sqrt( x + y ) 2

27. with each term is associated a fuzzy set
Fuzzy Set Model
Queries and docs represented by sets of index terms: matching is approximate from the start
This vagueness can be modeled using a fuzzy framework, as follows:
with each term is associated a fuzzy set
each doc has a degree of membership in this fuzzy set
This interpretation provides the foundation for many models for IR based on fuzzy theory

28. Probabilistic Model P(REL|D)
Views retrieval as an attempt to answer a basic question: “What is the probability that this document is relevant to this query?”
expressed as:
P(REL|D)
ie. Probability of x given y (Probability that of relevance given a particular document D)

29. Probabilistic Model An initial set of documents is retrieved somehow
User inspects these docs looking for the relevant ones (in truth, only top need to be inspected)
The system uses this information to refine description of ideal answer set
By repeting this process, it is expected that the description of the ideal answer set will improve
Have always in mind the need to guess at the very beginning the description of the ideal answer set
Description of ideal answer set is modeled in probabilistic terms

30. Recombination after dimensionality reduction

31. Classic IR Models Vector vs. probabilistic
“Numerous experiments demonstrate that probabilistic retrieval procedures yield good results. However, the results have not been sufficiently better than those obtained using Boolean or vector techniques to convince system developers to move heavily in this direction

32. Example Build the inverted file for the following document
F1={Written Quiz for Algorithms and Techniques of Information Retrieval}
F2={Program Quiz for Algorithms and Techniques of Web Search}
F3={Search on the Web for Information on Algorithms}

33. Example
You have the collection of documents that contain the following index terms:
D1: alpha bravo charlie delta echo foxtrot golf
D2: golf golf golf delta alpha
D3: bravo charlie bravo echo foxtrot bravo
D4: foxtrot alpha alpha golf golf delta
Use a frequency matrix of terms to calculate a similarity matrix for these documents, with weights proportional to the term frequency and inversely proportional to the document frequency.

34. Terms Documents
c1 c2 c3 c4 c5 m1 m2 m3 m4
__ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __ __
human
interface
computer
user
system
response
time
EPS
survey
trees
graph
minors
Give the scores of the 9 documents for the query trees, minors using Boolean search
Give the scores of the 9 documents for the query trees, minors using the vector model.