Lecture 3: Dictionaries and Tolerant Retrieval

Information Retrieval and Web Analytics

• Recap on the previous lecture

• The type/token distinction:

• Terms are normalized types put in the dictionary
• Tokenization problems:
• Hyphens, apostrophes, compounds, CJK
• Term equivalence classing:
• Numbers, case folding, stemming, lemmatization
• Skip pointers
• Encoding a tree-like structure in a postings list
• Biword indexes for phrases
• Positional indexes for phrases/proximity queries

1
This lecture

Dictionary data structures

“Tolerant” retrieval
Wild-card queries
Spelling correction
Soundex – words with similar pronunciations

2
 Document Collection

Document Normalization
Indexer

Normalization
Query IR System
Query

Indexes
UI

Ranking/Matching module

Set of relevant documents

3
Type/token distinction
Token an instance of a word or term occurring in a document
Type an equivalence class of tokens

In June, the dog likes to chase the cat in the barn.

12 word tokens
9 word types

4
Problems with equivalence classing
A term is an equivalence class of tokens.
How do we define equivalence classes?
Numbers (3/20/91 vs. 20/3/91)
Case folding
Stemming, Porter stemmer
Morphological analysis: inflectional vs. derivational
Equivalence classing problems in other languages

5
Positional indexes
Postings lists in a non-positional index: each posting is just a docID
Postings lists in a positional index: each posting is a docID and a list of
positions
Example query: “to1 be2 or3 not4 to5 be6”
With a positional index, we can answer
phrase queries
proximity queries

6
 Document Collection

Document Normalization
Indexer

Normalization
Query IR System
Query

Indexes
UI

Ranking/Matching module

Set of relevant documents

7
Dictionary data structures for inverted
indexes
The dictionary data structure stores the term vocabulary, document
frequency, pointers to each postings list … in what data structure?

8
A naïve dictionary

An array of struct:

char[20] int Postings *

20 bytes 4/8 bytes 4/8 bytes
How do we store a dictionary in memory efficiently?
How do we quickly look up elements at query time?

9
Dictionary data structures
Two main choices:
Hash tables
Trees
Some IR systems use hash tables, some trees
How many keys are we likely to have?
Is the number likely to remain static, or change a lot – and in the case of
changes, are we likely to only have new keys inserted, or to also have some keys
in the dictionary be deleted
What are the relative frequencies with which various keys will be accessed

10
Hashtables
Each vocabulary term is hashed to an integer
(We assume you’ve seen hashtables before)
Pros:
Lookup is faster than for a tree: O(1)
Cons:
No easy way to find minor variants:
judgment/judgement
No prefix search [tolerant retrieval]
all terms beginning with the prefix automat
If vocabulary keeps growing, need to occasionally do the expensive
operation of rehashing everything

11
Tree: Binary tree
Partition vocabulary terms into two subtrees, those whose first letter
is between a and m, and the rest (actual terms stored in the leafs).
Anything that is on the left subtree is smaller than what’s on the
right. Trees solve the prefix problem (find all terms starting with
automat).

12
Binary tree
Root
a-m n-z

a-hu hy-m n-sh si-z

13
Binary Tree
Cost of operations depends on height of tree.
Keep height minimum / keep binary tree balanced: for each node,
heights of subtrees differ by no more than 1. O(log M) search for
balanced trees, where M is the size of the vocabulary.
Search is slightly slower than in hashes But: re-balancing binary trees
is expensive (insertion and deletion of terms).

14
Tree: B-tree

a-hu n-z
hy-m

Definition: Every internal nodel has a number of children in the interval [a,b]
where a, b are appropriate natural numbers, e.g., [2,4].

15
Trees
Simplest: binary tree
More usual: B-trees
Trees require a standard ordering of characters and hence
strings … but we typically have one
Pros:
Solves the prefix problem (terms starting with hyp)

Cons:
Slower: O(log M) [and this requires balanced tree]
Rebalancing binary trees is expensive
But B-trees mitigate the rebalancing problem

16
Wild-card queries
Why do we need wild-card queries
• the user is uncertain of the spelling of a query term. (e.g., Sydney vs.
Sidney).
• This leads to a wild-card query S*dney
• the user is aware of multiple variants of spelling a term. (e.g., color vs.
colour).
• the user seeks documents containing variants of a term that would be
caught by stemming, (e.g., judicial vs. judiciary, leading to the wildcard
query judicia*)
• the user is uncertain of the correct rendition of a foreign word or phrase
(e.g., the query Universit* Stuttgart

17
Wild-card queries: *

mon*: find all docs containing any word beginning with “mon”.
Easy with binary tree (or B-tree) lexicon: retrieve all words in
range: mon ≤ w < moo
*mon: find words ending in “mon”: harder
Maintain an additional B-tree for terms backwards (reverse B-
tree).
Then retrieve all terms t in subtree rooted at n-o-m

Exercise: from this, how can we enumerate all terms

meeting the wild-card query pro*cent ?
18
Query processing
At this point, we have a list of all terms in the dictionary that match
the wild-card query.
We still have to look up the postings for each enumerated term.
E.g., consider the query:
se*ate AND fil*er
This may result in the execution of many Boolean AND queries.`

19
B-trees handle *’s at the end of a query
term
How can we handle *’s in the middle of query term?
co*tion
We could look up co* AND *tion in a B-tree and intersect the two term
sets
Expensive
The solution: transform wild-card queries so that the *’s occur at the
end
This gives rise to the Permuterm Index.

20
Permuterm index
For term hello, index under:
hello$, ello$h, llo$he, lo$hel, o$hell, $hello
where $ is a special symbol.
Permuterm Dictionary/Vocabulary

Postings: Every posting

hello$ list has only a single entry
which is original term of
the rotated index
$hello

Permuterm o$hell hello

vocabulary/
Dictionary
lo$hel

llo$he 21

Ello$h
 X=>String of 1 or
more characters
Queries:
X lookup on X$ X* lookup on $X*
*X lookup on X$* *X* lookup on X*
Query = hel*o
X*Y lookup on Y$X* fi*mo*er???
X=hel, Y=o
er$fi* -> fishmonger & filibuster
Lookup o$hel*
Y$X*???
Look up for y$x prefix in dictionary of the permuterm index and get all the
corresponding terms in the posting list for all those and collect them together.
Then by taking all those original terms and check them in standard
inverted index to obtain Doc ids of those terms and get union of
them.

22
Permuterm query processing

Rotate query wild-card to the right

Now use B-tree lookup as before.
Permuterm problem: ≈ quadruples lexicon size

Empirical observation for English.

23
k-gram indexes

Enumerate all k-grams (sequence of k chars) occurring in any term

e.g., from text “April is the cruelest month” we get the 2-grams
(bigrams)
$a,ap,pr,ri,il,l$,$i,is,s$,$t,th,he,e$,$c,cr,ru,
ue,el,le,es,st,t$, $m,mo,on,nt,h$

$ is a special word boundary symbol

Maintain a second inverted index from bigrams to dictionary terms
that match each bigram.

24
Bigram index example
The k-gram index finds terms based on a query consisting
of k-grams (here k=2).

$m mace madden

mo among am ortize
on along among

25
Processing wild-cards
Query mon* can now be run as
$m AND mo AND on
Gets terms that match AND version of our wildcard query.
But we’d enumerate moon. False positive
Must post-filter these terms against query.
Surviving enumerated terms are then looked up in the term-document inverted
index.
Fast, space efficient (compared to permuterm). (Since we are not storing every
possible rotations of a single term)
k-gram vs. permuterm index
k-gram index is more space-efficient
permuterm index does not require postfiltering always (only need for some
cases with multiple wild cards).
26
Processing wild-card queries
As before, we must execute a Boolean query for each enumerated,
filtered term.
Wild-cards can result in expensive query execution (very large
disjunctions…)
pyth* AND prog*
If you encourage “laziness” people will respond!

Search

Type your search terms, use ‘*’ if you need to.

E.g., Alex* will match Alexander.

Which web search engines allow wildcard queries?

27
Spell correction
Two principal uses
Correcting document(s) being indexed
Correcting user queries to retrieve “right” answers
Two main methods:
Isolated word
Check each word on its own for misspelling
Return the “correct” word that has the smallest distance to the misspelled word.
Will not catch typos resulting in correctly spelled words
e.g., from → form

Context-sensitive
Look at surrounding words,
e.g., I flew form Heathrow to Narita.

28
Document correction
Especially needed for OCR’ed documents
Correction algorithms are tuned for this: r n/m
Can use domain-specific knowledge
E.g., OCR can confuse O and D more often than it would confuse O
and I (adjacent on the QWERTY keyboard, so more likely interchanged
in typing).

But also: web pages and even printed material have typos
Goal: the dictionary contains fewer misspellings
But often we don’t change the documents and instead fix
the query-document mapping

29
Query mis-spellings
Our principal focus here
E.g., the query Alanis Morisett
We can either
Retrieve documents indexed by the correct spelling, OR
Return several suggested alternative queries with the correct
spelling
Did you mean … ?

30
Isolated word correction
Fundamental premise – there is a lexicon from which the
correct spellings come
Two basic choices for this
A standard lexicon such as
Webster’s English Dictionary
An “industry-specific” lexicon – hand-maintained
The lexicon of the indexed corpus
E.g., all words on the web
All names, acronyms etc.
(Including the mis-spellings)

31
Isolated word correction
There is a list of “correct” words – for instance a standard dictionary
(Webster’s, OED. . . )
Return the “correct” word that has the smallest distance to the
misspelled word.
Then we need a way of computing the distance between a
misspelled word and a correct word
Edit distance (Levenshtein distance)
Weighted edit distance
n-gram overlap

32
Edit distance (Levenshtein distance)
Given two strings S1 and S2, the minimum number of operations to
convert one to the other
Operations are typically character-level
Insert, Delete, Replace, (Transposition)
E.g., the edit distance from dof to dog is 1
From cat to act is 2 (Just 1 with transpose.)
from cat to dog is 3.
Generally found by dynamic programming.

33
Edit distance (Levenshtein distance)

Edit distance between two strings s1 and s2 is the minimum number of basic
operations that transform s1 into s2.
Levenshtein distance: Acceptable operations are insert, delete and replace

horse ros Operations

1. REPLACE
horse rorse ( REPLACE ‘h’ with ‘r’) 2. INSERT
3. DELETE
rorse rose ( DELETE ‘r’)

rose ros ( DELETE ‘e’)

Edit distance = 3 34
Weighted edit distance

As above, but the weight of an operation depends on the

character(s) involved
Meant to capture OCR or keyboard errors
Example: m more likely to be mis-typed as n than as q
Therefore, replacing m by n is a smaller edit distance than by q
This may be formulated as a probability model
Requires weight matrix as input
Modify dynamic programming to handle weights

35
Edit distance to all dictionary terms?

Given a (mis-spelled) query – do we compute its edit distance to every

dictionary term?
Expensive and slow
Alternative?
How do we cut the set of candidate dictionary terms?
One possibility is to use n-gram overlap for this
This can also be used by itself for spelling correction.

36
n-gram overlap

Enumerate all the n-grams in the query string as well as in the

lexicon
Use the n-gram index (recall wild-card search) to retrieve all
lexicon terms matching any of the query n-grams
Threshold by number of matching n-grams
Variants – weight by keyboard layout, etc.

37
Example with trigrams

Suppose the text is november

Trigrams are nov, ove, vem, emb, mbe, ber.
The query is december
Trigrams are dec, ece, cem, emb, mbe, ber.
So 3 trigrams overlap (of 6 in each term)
How can we turn this into a normalized measure of overlap?

38
One option – Jaccard coefficient

A commonly-used measure of overlap

Let X and Y be two sets; then the J.C. is

X Y / X Y
Equals 1 when X and Y have the same elements and zero when they
are disjoint
X and Y don’t have to be of the same size
Always assigns a number between 0 and 1
Now threshold to decide if you have a match
E.g., if J.C. > 0.8, declare a match

39
Another option
Consider the query lord – we wish to identify words matching 2 of its 3
bigrams (lo, or, rd)

lo alone lore sloth

or border lore morbid
rd ardent border card

Standard postings “merge” will enumerate …

Adapt this to using Jaccard (or another) mea4s0 ure.
Context-sensitive spell correction

Text: I flew from Heathrow to Narita.

Consider the phrase query “flew form Heathrow”
We’d like to respond
Did you mean “flew from Heathrow”?
because no docs matched the query phrase.

41
Context-sensitive correction

Need surrounding context to catch this.

First idea: retrieve dictionary terms close (in weighted edit distance)
to each query term
Now try all possible resulting phrases with one word “fixed” at a time
flew from heathrow
fled form heathrow
flea form heathrow
Hit-based spelling correction: Suggest the alternative that has lots of
hits.

42
Exercise

Suppose that for “flew form Heathrow” we have 7 alternatives for flew, 19
for form and 3 for heathrow.
How many “corrected” phrases will we enumerate in this scheme?

43
Another approach

Biword Indexing we discussed in the Lecture 02

44
Soundex

45
Soundex
Class of heuristics to expand a query into phonetic equivalents
Language specific – mainly for names
E.g., chebyshev → tchebycheff
Invented for the U.S. census … in 1918

46
Soundex – typical algorithm

Turn every token to be indexed into a 4-character reduced form

Do the same with query terms
Build and search an index on the reduced forms
(when the query calls for a soundex match)

47
Soundex – typical algorithm

1. Retain the first letter of the word.

2. Change all occurrences of the following letters to '0' (zero):
'A', E', 'I', 'O', 'U', 'H', 'W', 'Y'.
3. Change letters to digits as follows:
B, F, P, V → 1
C, G, J, K, Q, S, X, Z → 2
D,T → 3
L→ 4
M, N → 5
R→6 48
Soundex continued
4. Remove all pairs of consecutive digits.
5. Remove all zeros from the resulting string.
6. Pad the resulting string with trailing zeros and return the first four positions,
which will be of the form <uppercase letter> <digit> <digit> <digit>.

E.g., Herman becomes H655.

Will hermann generate the same code?

49
Soundex

Soundex is the classic algorithm, provided by most databases (Oracle,

Microsoft, …)
How useful is soundex?
Not very – for information retrieval
Okay for “high recall” tasks (e.g., Interpol), though biased to names
of certain nationalities
Zobel and Dart (1996) show that other algorithms for phonetic
matching perform much better in the context of IR

50
What queries can we process?

We have
Positional inverted index with skip pointers
Wild-card index
Spell-correction
Soundex
Queries such as
(SPELL(moriset) /3 toron*to) OR SOUNDEX(chaikofski)

51
Exercise

Draw yourself a diagram showing the various indexes in a search

engine incorporating all the functionality we have talked about
Identify some of the key design choices in the index pipeline:
Does stemming happen before the Soundex index?
What about n-grams?
Given a query, how would you parse and dispatch sub-queries to the
various indexes?

52
Resources

IIR 3, MG 4.2
Efficient spell retrieval:
K. Kukich. Techniques for automatically correcting words in text. ACM
Computing Surveys 24(4), Dec 1992.
J. Zobel and P. Dart. Finding approximate matches in large
lexicons. Software - practice and experience 25(3), March 1995.
http://citeseer.ist.psu.edu/zobel95finding.html
Mikael Tillenius: Efficient Generation and Ranking of Spelling Error
Corrections. Master’s thesis at Sweden’s Royal Institute of Technology.
http://citeseer.ist.psu.edu/179155.html
Nice, easy reading on spell correction:
Peter Norvig: How to write a spelling corrector
http://norvig.com/spell-correct.html
53