1. chapter 1 I N F O R M A T I O N R E T R I E V A L & VISUALIZATION

2. Outline What is the IR problem?
How to organize an IR system? (Or the main processes in IR)
Tokenizing
Indexing
Retrieval
System evaluation
Some current research topics

3. Information Retrieval
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. IR vs. databases: Structured vs unstructured data
Structured data tends to refer to information in “tables”
Employee
Manager
Salary
Smith
Jones
50000
Chang
Smith
60000
Ivy
Smith
50000
Typically allows numerical range and exact match
(for text) queries, e.g.,
Salary < AND Manager = Smith. 4

5. Unstructured data Typically refers to free text Allows
Keyword queries including operators
More sophisticated “concept” queries e.g.,
find all web pages dealing with drug abuse 5

6. Unstructured (text) vs. structured (database) data in 1996

7. Unstructured (text) vs. structured (database) data in 2009

8. IR problem First applications: in libraries (1950s)
ISBN:
Author: Salton, Gerard
Title: Automatic text processing: the transformation, analysis, and retrieval of information by computer
Editor: Addison-Wesley
Date: 1989
Content: <Text>
external attributes and internal attribute (content)
Search by external attributes = Search in DB
IR: search by content

9. The problem of IR
Goal = find documents relevant to an information need from a large document set
Info. need
Query
IR system
Document
collection
Retrieval
Answer list

10. Example
Google
Web

11. The process of retrieving information

12. The process of retrieving information
Generating logical view of documents.
Building an index of the text database.
Initiating retrieval process by specifying a user need.
Generating logical view of user need (query).
Retrieving documents by Processing the query.
Ranking documents upon relevance score.
Initiating a user feedback cycle
(modifying user need _modifying query).

13. Basic assumptions of Information Retrieval
Collection: Fixed set of documents
Goal: Retrieve documents with information that is relevant to the user’s information need and helps the user complete a task

14. Generating logical view of documents Inverted index construction
Documents to
be indexed.
Friends, Romans, countrymen.
Tokenizer
Token stream.
Friends
Romans
Countrymen
More on
these later.
Linguistic modules
Modified tokens.
friend
roman
countryman
Indexer
Inverted index.
friend
roman
countryman 2 4 13 16 1

15. Generating logical view of documents Parsing a document
What format is it in?
pdf/word/excel/html?
What language is it in?
What character set is in use?
Each of these is a classification problem.
But these tasks are often done heuristically …

16. Generating logical view of documents Tokenization
Input: “Friends, Romans and Countrymen”
Output: Tokens
Friends
Romans
Countrymen
A token is an instance of a sequence of characters
Each such token is now a candidate for an index entry, after further processing
Described below
But what are valid tokens to emit?

17. Generating logical view of documents Tokenization
Stop words
With a stop list, you exclude from the dictionary entirely the commonest words. Intuition:
They have little semantic content: the, a, and, to, be…
Normalization to terms
We need to “normalize” words in indexed text as well as query words into the same form
We want to match U.S.A. and USA

18. Generating logical view of documents Tokenization
Case folding
Reduce all letters to lower case
exception: upper case in mid-sentence?
e.g., General Motors
Fed vs. fed
SAIL vs. sail
Thesauri and soundex
car = automobile color = colour
What about spelling mistakes?
One approach is soundex, which forms equivalence classes of words based on phonetic heuristics

19. Generating logical view of documents Tokenizing
Lemmatization
Reduce inflectional/variant forms to base form
E.g.,
am, are, is  be
car, cars, car's, cars'  car
the boy's cars are different colors  the boy car be different color

20. Generating logical view of documents Tokenizing
Stemming
Reduce terms to their “roots” before indexing
“Stemming” suggest crude affix chopping
language dependent
e.g., automate(s), automatic, automation all reduced to
automat.
for exampl compress and
compress ar both accept
as equival to compress
for example compressed
and compression are both
accepted as equivalent to
compress.

21. Generating logical view of documents Tokenizing
Porter’s algorithm
Commonest algorithm for stemming English
sses  ss
ies  i
ational  ate
tional  tion
Weight of word sensitive rules
(m>1) EMENT →
replacement → replac
cement → cement

22. Building an index of the text database Term-document incidence –Matrix Dictionary
1 if play contains word, 0 otherwise
Brutus AND Caesar BUT NOT Calpurnia

23. Building an index of the text database Bigger collections
Consider N = 1 million documents, each with about 1000 words.
Avg 6 bytes/word including spaces/punctuation
6GB of data in the documents.
Say there are M = 500K distinct terms among these.

24. Building an index of the text database Can’t build the matrix
500K x 1M matrix has half-a-trillion 0’s and 1’s.
But it has no more than one billion 1’s.
matrix is extremely sparse.
What’s a better representation?
We only record the 1 positions.
Why?

25. Building an index of the text database Inverted index
For each term t, we must store a list of all documents that contain t.
Identify each by a docID, a document serial number
Can we used fixed-size arrays for this?
Brutus 1 2 4 11 31 45
173
174
Caesar 1 2 4 5 6 16 57
132
Calpurnia 2 31 54
101
What happens if the word Caesar is added to document 14?

26. Building an index of the text database Inverted index
We need variable-size postings lists
On disk, a continuous run of postings is normal and best
In memory, can use linked lists or variable length arrays
Some tradeoffs in size/ease of insertion
Posting
Brutus 1 2 4 11 31 45
173
174
Dictionary
Caesar 1 2 4 5 6 16 57
132
Linked lists generally preferred to arrays
Dynamic space allocation
Insertion of terms into documents easy
Space overhead of pointers
Calpurnia 2 31 54
101
Postings
Sorted by docID (more later on why).

27. Building an index of the text database Indexer steps: Token sequence
Sequence of (Modified token, Document ID) pairs.
Doc 1
Doc 2
I did enact Julius
Caesar I was killed
i' the Capitol;
Brutus killed me.
So let it be with
Caesar. The noble
Brutus hath told you
Caesar was ambitious

28. Building an index of the text database Indexer steps: Sort
Sort by terms
And then docID
Core indexing step

29. Building an index of the text database Indexer steps: Dictionary & Postings
Multiple term entries in a single document are merged.
Split into Dictionary and Postings
Doc. frequency information is added.
Why frequency?
Will discuss later.

30. Building an index of the text database Where do we pay in storage?
Lists of docIDs
Terms and counts
How do we index efficiently?
How much storage do we need?
Pointers

31. An example

32. Retrieval The index we just built
How do we process a query?
what kinds of queries can we process?
Retrieving

33. Term-document incidence
1 if play contains word, 0 otherwise
Brutus AND Caesar BUT NOT Calpurnia

34. Incidence vectors So we have a 0/1 vector for each term.
To answer query: take the vectors for Brutus, Caesar and Calpurnia (complemented)  bitwise AND.
AND AND =
1 represent the play is ok satisfy the condition
Here play 1,4 satisfy the query (antony and cleopatra and hamlet) 34

35. Answers to query Antony and Cleopatra, Act III, Scene ii
Agrippa [Aside to DOMITIUS ENOBARBUS]: Why, Enobarbus,
When Antony found Julius Caesar dead,
He cried almost to roaring; and he wept
When at Philippi he found Brutus slain.
Hamlet, Act III, Scene ii
Lord Polonius: I did enact Julius Caesar I was killed i' the
Capitol; Brutus killed me. 35

36. Query processing: AND Consider processing the query: Brutus AND Caesar
Locate Brutus in the Dictionary;
Retrieve its postings.
Locate Caesar in the Dictionary;
“Merge” the two postings: 2 4 8 16 32 64
128
Brutus 1 2 3 5 8 13 21 34
Caesar

37. The merge
Walk through the two postings simultaneously, in time linear in the total number of postings entries 34
128 2 4 8 16 32 64 1 3 5 13 21 2 4 8 16 32 64
128
Brutus
Caesar 2 8 1 2 3 5 8 13 21 34
If the list lengths are x and y, the merge takes O(x+y)
operations.
Crucial: postings sorted by docID.

38. Boolean queries: Exact match
The Boolean retrieval model is being able to ask a query that is a Boolean expression:
Boolean Queries are queries using AND, OR and NOT to join query terms
Views each document as a set of words
Is precise: document matches condition or not.
Perhaps the simplest model to build an IR system on
Primary commercial retrieval tool for 3 decades.
Many search systems you still use are Boolean:
, library catalog, Mac OS X Spotlight

39. Ranking search results
Boolean queries give inclusion or exclusion of docs.
Often we want to rank/group results
Need to measure proximity from query to each doc.
Need to decide whether docs presented to user are singletons, or a group of docs covering various aspects of the query.

40. Ranked retrieval Thus far, our queries have all been Boolean.
Documents either match or don’t.
Good for expert users with precise understanding of their needs and the collection.
Also good for applications: Applications can easily consume 1000s of results.
Not good for the majority of users.
Most users incapable of writing Boolean queries (or they are, but they think it’s too much work).
Most users don’t want to wade through 1000s of results.
This is particularly true of web search.

41. Ranked retrieval models
Rather than a set of documents satisfying a query expression, in ranked retrieval models, the system returns an ordering over the (top) documents in the collection with respect to a query
Free text queries: Rather than a query language of operators and expressions, the user’s query is just one or more words in a human language
In principle, there are two separate choices here, but in practice, ranked retrieval models have normally been associated with free text queries and vice versa 41

42. Feast or famine: not a problem in ranked retrieval
When a system produces a ranked result set, large result sets are not an issue
Indeed, the size of the result set is not an issue
We just show the top k ( ≈ 10) results
We don’t overwhelm the user
Premise: the ranking algorithm works

43. Scoring as the basis of ranked retrieval
We wish to return in order the documents most likely to be useful to the searcher
How can we rank-order the documents in the collection with respect to a query?
Assign a score – say in [0, 1] – to each document
This score measures how well document and query “match”.

44. Query-document matching scores
We need a way of assigning a score to a query/document pair
Let’s start with a one-term query
If the query term does not occur in the document: score should be 0
The more frequent the query term in the document, the higher the score (should be)
We will look at a number of alternatives for this.

45. Take 1: Jaccard coefficient
A commonly used measure of overlap of two sets A and B
jaccard(A,B) = |A ∩ B| / |A ∪ B|
jaccard(A,A) = 1
jaccard(A,B) = 0 if A ∩ B = 0
A and B don’t have to be the same size.
Always assigns a number between 0 and 1.

46. Issues with Jaccard for scoring
It doesn’t consider term frequency (how many times a term occurs in a document)
Rare terms in a collection are more informative than frequent terms. Jaccard doesn’t consider this information
We need a more sophisticated way of normalizing for length
Later , we’ll use
. . . instead of |A ∩ B|/|A ∪ B| (Jaccard) for length normalization.

47. Binary term-document incidence matrix
Each document is represented by a binary vector ∈ {0,1}|V|

48. Term-document count matrices
Consider the number of occurrences of a term in a document:
Each document is a count vector in ℕv: a column below

49. Bag of words model
Vector representation doesn’t consider the ordering of words in a document
John is quicker than Mary and Mary is quicker than John have the same vectors
This is called the bag of words model.
In a sense, this is a step back: The positional index was able to distinguish these two documents.
We will look at “recovering” positional information later .
For now: bag of words model

50. NB: frequency = count in IR
Term frequency tf
The term frequency tft,d of term t in document d is defined as the number of times that t occurs in d.
We want to use tf when computing query-document match scores. But how?
Raw term frequency is not what we want:
A document with 10 occurrences of the term is more relevant than a document with 1 occurrence of the term.
But not 10 times more relevant.
Relevance does not increase proportionally with term frequency.
NB: frequency = count in IR

51. Log-frequency weighting
The log frequency weight of term t in d is
0 → 0, 1 → 1, 2 → 1.3, 10 → 2, 1000 → 4, etc.
Score for a document-query pair: sum over terms t in both q and d:
score
The score is 0 if none of the query terms is present in the document.

52. Document frequency Rare terms are more informative than frequent terms
Recall stop words
Consider a term in the query that is rare in the collection (e.g., arachnocentric)
A document containing this term is very likely to be relevant to the query arachnocentric
→ We want a high weight for rare terms like arachnocentric.

53. Document frequency, continued
Frequent terms are less informative than rare terms
Consider a query term that is frequent in the collection (e.g., high, increase, line)
A document containing such a term is more likely to be relevant than a document that doesn’t
But it’s not a sure indicator of relevance.
→ For frequent terms, we want high positive weights for words like high, increase, and line
But lower weights than for rare terms.
We will use document frequency (df) to capture this.

54. idf weight
dft is the document frequency of t: the number of documents that contain t
dft is an inverse measure of the informativeness of t
dft  N
We define the idf (inverse document frequency) of t by
We use log (N/dft) instead of N/dft to “dampen” the effect of idf.
Will turn out the base of the log is immaterial.

55. idf example, suppose N = 1 million
term
dft
idft
calpurnia 1
animal
100
sunday
1,000
fly
10,000
under
100,000
the
1,000,000
There is one idf value for each term t in a collection. 55

56. Effect of idf on ranking
Does idf have an effect on ranking for one-term queries, like
iPhone
idf has no effect on ranking one term queries
idf affects the ranking of documents for queries with at least two terms
For the query capricious person, idf weighting makes occurrences of capricious count for much more in the final document ranking than occurrences of person. 56

57. Collection vs. Document frequency
The collection frequency of t is the number of occurrences of t in the collection, counting multiple occurrences.
Example:
Which word is a better search term (and should get a higher weight)?
Word
Collection frequency
Document frequency
insurance
10440
3997
try
10422
8760
Why do you get these numbers?
Suggests df is better. 57

58. tf-idf weighting
The tf-idf weight of a term is the product of its tf weight and its idf weight.
Best known weighting scheme in information retrieval
Note: the “-” in tf-idf is a hyphen, not a minus sign!
Alternative names: tf.idf, tf x idf
Increases with the number of occurrences within a document
Increases with the rarity of the term in the collection

59. Final ranking of documents for a query59

60. Binary → count → weight matrix
Each document is now represented by a real-valued vector of tf-idf weights ∈ R|V|

61. Documents as vectors So we have a |V|-dimensional vector space
Terms are axes of the space
Documents are points or vectors in this space
Very high-dimensional: tens of millions of dimensions when you apply this to a web search engine
These are very sparse vectors - most entries are zero.

62. Queries as vectors
Key idea 1: Do the same for queries: represent them as vectors in the space
Key idea 2: Rank documents according to their proximity to the query in this space
proximity = similarity of vectors
proximity ≈ inverse of distance
Recall: We do this because we want to get away from the you’re-either-in-or-out Boolean model.
Instead: rank more relevant documents higher than less relevant documents

63. Formalizing vector space proximity
Sec. 6.3
Formalizing vector space proximity
First cut: distance between two points
( = distance between the end points of the two vectors)
Euclidean distance?
Euclidean distance is a bad idea . . .
. . . because Euclidean distance is large for vectors of different lengths.

64. Why distance is a bad idea
Sec. 6.3
Why distance is a bad idea
The Euclidean distance between q
and d2 is large even though the
distribution of terms in the query q and the distribution of
terms in the document d2 are
very similar.

65. Use angle instead of distance
Sec. 6.3
Use angle instead of distance
Thought experiment: take a document d and append it to itself. Call this document d′.
“Semantically” d and d′ have the same content
The Euclidean distance between the two documents can be quite large
The angle between the two documents is 0, corresponding to maximal similarity.
Key idea: Rank documents according to angle with query.

66. From angles to cosines The following two notions are equivalent.
Sec. 6.3
From angles to cosines
The following two notions are equivalent.
Rank documents in increasing g order of the angle between query and document
Rank documents in decreasing order of cosine(query,document)
Cosine is a monotonically decreasing function for the interval [0o, 180o]

67. Sec. 6.3
From angles to cosines
But how – and why – should we be computing cosines?

68. Sec. 6.3
Length normalization
A vector can be (length-) normalized by dividing each of its components by its length – for this we use the L2 norm:
Dividing a vector by its L2 norm makes it a unit (length) vector (on surface of unit hypersphere)
Effect on the two documents d and d′ (d appended to itself) from earlier slide: they have identical vectors after length-normalization.
Long and short documents now have comparable weights

69. cosine(query,document)
Dot product
Unit vectors
qi is the tf-idf weight of term i in the query
di is the tf-idf weight of term i in the document
cos(q,d) is the cosine similarity of q and d … or,
equivalently, the cosine of the angle between q and d.
See Law of Cosines (Cosine Rule) wikipedia page 69

70. Cosine for length-normalized vectors
For length-normalized vectors, cosine similarity is simply the dot product (or scalar product):
for q, d length-normalized. 70

71. Cosine similarity illustrated

72. Cosine similarity amongst 3 documents
How similar are
the novels
SaS: Sense and
Sensibility
PaP: Pride and
Prejudice, and
WH: Wuthering
Heights?
term
SaS
PaP WH
affection
115 58 20
jealous 10 7 11
gossip 2 6
wuthering 38
Term frequencies (counts)
Note: To simplify this example, we don’t do idf weighting.

73. 3 documents example contd.
Log frequency weighting
After length normalization
term
SaS
PaP WH
affection
3.06
2.76
2.30
jealous
2.00
1.85
2.04
gossip
1.30
1.78
wuthering
2.58
term
SaS
PaP WH
affection
0.789
0.832
0.524
jealous
0.515
0.555
0.465
gossip
0.335
0.405
wuthering
0.588
Now length normalizion for word “affection” =
3.06/
cos(SaS,PaP) ≈
0.789 × × × × 0.0 ≈ 0.94
cos(SaS,WH) ≈ 0.79
cos(PaP,WH) ≈ 0.69
Why do we have cos(SaS,PaP) > cos(SAS,WH)?

74. Computing cosine scores

75. tf-idf weighting has many variants
n default is just term frequency
ltc is best known form of weighting
Columns headed ‘n’ are acronyms for weight schemes.
Why is the base of the log in idf immaterial? 75

76. Weighting may differ in queries vs documents
Sec. 6.4
Weighting may differ in queries vs documents
Many search engines allow for different weightings for queries vs. documents
SMART Notation: denotes the combination in use in an engine, with the notation ddd.qqq, using the acronyms from the previous table
A very standard weighting scheme is: lnc.ltc
Document: logarithmic tf (l as first character), no idf and cosine normalization
Query: logarithmic tf (l in leftmost column), idf (t in second column), no normalization …
Leaving off idf weighting on documents is good for both efficiency and system effectiveness reasons.
A bad idea?

77. tf-idf example: lnc.ltc
Document: car insurance auto insurance
Query: best car insurance
Term
Query
Document
Prod
tf-raw
tf-wt df
idf wt
n’lize
auto
5000
2.3 1
0.52
best
50000
1.3
0.34
car
10000
2.0
0.27
insurance
1000
3.0
0.78 2
0.68
0.53
1+log (1)
Log(1,000,000/1000)
1x3
3/Sqr( )
1+log (2)
1.3/
(sqrt( )
=0.78x0.68
Exercise: what is N, the number of docs? N = 1,000,000
Doc length =
Score = = 0.8

78. How good are the retrieved docs?
Precision : Fraction of retrieved docs that are relevant to user’s information need
Recall : Fraction of relevant docs in collection that are retrieved 78

79. Unranked retrieval evaluation: Precision and Recall
Sec. 8.3
Unranked retrieval evaluation: Precision and Recall
Precision: fraction of retrieved docs that are relevant = P(relevant|retrieved)
Recall: fraction of relevant docs that are retrieved = P(retrieved|relevant)
Precision P = tp/(tp + fp)
Recall R = tp/(tp + fn)
Relevant
Nonrelevant
Retrieved tp fp
Not Retrieved fn tn

80. Rank precision
Compute the precision values at some selected rank positions.
Mainly used in Web search evaluation.
For a Web search engine, we can compute precisions for the top 5, 10, 15, 20, 25 and 30 returned pages
as the user seldom looks at more than 30 pages.
Recall is not very meaningful in Web search.
Why?
CS583, Bing Liu, UIC

81. Should we instead use the accuracy measure for evaluation?
Sec. 8.3
Should we instead use the accuracy measure for evaluation?
Given a query, an engine classifies each doc as “Relevant” or “Nonrelevant”
The accuracy of an engine: the fraction of these classifications that are correct
(tp + tn) / ( tp + fp + fn + tn)
Accuracy is a commonly used evaluation measure in machine learning classification work
Why is this not a very useful evaluation measure in IR?

82. Why not just use accuracy?
Sec. 8.3
Why not just use accuracy?
How to build a % accurate search engine on a low budget….
People doing information retrieval want to find something and have a certain tolerance for junk.
Snoogle.com
Search for:
0 matching results found.

83. Sec. 8.3
Precision/Recall
You can get high recall (but low precision) by retrieving all docs for all queries!
Recall is a non-decreasing function of the number of docs retrieved
In a good system, precision decreases as either the number of docs retrieved or recall increases
This is not a theorem, but a result with strong empirical confirmation

84. Sec. 8.3
A combined measure: F
Combined measure that assesses precision/recall tradeoff is F measure (weighted harmonic mean):
People usually use balanced F1 measure
i.e., with  = 1 or  = ½
Harmonic mean is a conservative average
See CJ van Rijsbergen, Information Retrieval

85. Evaluating ranked results
Sec. 8.4
Evaluating ranked results
Evaluation of ranked results:
The system can return any number of results
By taking various numbers of the top returned documents (levels of recall), the evaluator can produce a precision-recall curve

86. User Feed Back Initiating a user feedback cycle depending on results
modifying user need modifying query.

87. Relevance feedback
Relevance feedback is one of the techniques for improving retrieval effectiveness. The steps:
the user first identifies some relevant (Dr) and irrelevant documents (Dir) in the initial list of retrieved documents
the system expands the query q by extracting some additional terms from the sample relevant and irrelevant documents to produce qe
Perform a second round of retrieval.
Rocchio method (α, β and γ are parameters)
CS583, Bing Liu, UIC

88. Rocchio text classifier
In fact, a variation of the Rocchio method above, called the Rocchio classification method, can be used to improve retrieval effectiveness too
so are other machine learning methods. Why?
Rocchio classifier is constructed by producing a prototype vector ci for each class i (relevant or irrelevant in this case):
In classification, cosine is used.
CS583, Bing Liu, UIC

89. Some current research topics
a bag-of-words model
While term frequency is certainly one of the most useful indicators of relevance, relying solely on term frequency amounts to viewing a document simply as a bag-of-words and ignores much of its structure.

90. Some current research topics
information derived from document structure
Term order is an example of structural information that some recent approaches utilize.
Such approaches assign higher relevance scores to documents in which query terms appear in the same order as they appear in the query

91. Some current research topics
Document-feature based approaches take into account the occurrence of query terms in locations of varying prominence, such as title, header text, or body text.
The most prominent document-feature based approach is BM25F, built upon Okapi BM25 . Document feature techniques are most commonly used in web search as features can be easily identified due to the presence of HTML tags

92. Some current research topics
Term proximity,
is an idea which has seen recent interest.
In this context, term proximity refers to the lexical distance between query terms, calculated as the number of words separating query terms in a document

93. Some current research topics
Chronological Term Rank
The chronological rank of a term (CTR) within a document is defined as the rank of the term in the sequence of words in the original document. We refer to this statistic as “chronological” to emphasize its correspondence to the sequential occurrence of the terms within the document from the beginning to end
let D = (t1, , tn) be a document where ti are terms (words) ordered according to their sequence in the original document. Let trt := i, where the chronological rank tr of term t is assigned as the subscript i of the earliest occurrence of t in D.

94. Index construction
Easy! See the example,
CS583, Bing Liu, UIC

95. Different search engines
The real differences among different search engines are
their index weighting schemes
Including location of terms, e.g., title, body, emphasized words, etc.
their query processing methods (e.g., query classification, expansion, etc)
their ranking algorithms
Few of these are published by any of the search engine companies. They aretightly guarded secrets.
CS583, Bing Liu, UIC