SlideShare a Scribd company logo

information retrival and text processing

M
mausamraushan2288

The document discusses various aspects of information retrieval and text processing, including the indexing of documents, the importance of word frequency statistics, and the application of Zipf's Law in understanding word distributions. It also covers techniques such as tokenization, stemming, and link analysis that enhance search engine effectiveness by addressing challenges in document query matching and retrieval algorithms. Key models and methods, including the Random Surfer model and Pagerank, are highlighted for their role in estimating result set sizes and assessing web page importance.

Education
1 of 63
Download to read offline
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
In association with and Chapter 4 Processing Text Search Engines and Information Retrieval Full Credit: Croft et al. - http://www.search-engines-book.com/
In association with and From Ch. 2: Indexing Process
In association with and Processing Text ● Converting documents to index terms ● Why? ○ Matching the exact string of characters typed by the user is too restrictive ■ i.e., it doesn’t work very well in terms of effectiveness ■ e.g. uppercase/lowercase ○ Not all words are of equal value in a search ○ Sometimes not clear where words begin and end ■ Not even clear what a word is in some languages ● e.g., Chinese, Korean
In association with and Text Statistics ● Huge variety of words used in text but ● Many statistical characteristics of word occurrences are predictable ○ e.g., distribution of word counts ● Retrieval models and ranking algorithms depend heavily on statistical properties of words ○ e.g., important words occur often in documents but are not high frequency in collection Retrieval models are algorithms or frameworks used in information retrieval systems to match user queries with relevant documents from a collection of documents, such as web pages, articles, or books.
In association with and Zipf’s Law ● Distribution of word frequencies is very skewed ○ a few words occur very often, many words hardly ever occur ○ e.g., two most common words (“the”, “of”) make up about 10% of all word occurrences in text documents ● Zipf’s “law”: ○ observation that rank (r) of a word times its frequency (f) is approximately a constant (k) ■ assuming words are ranked in order of decreasing frequency ○ i.e., r.f ≈ k or r.Pr ≈ c, where Pr is probability of word occurrence and c ≈ 0.1 for English
In association with and Zipf’s Law
In association with and News Collection (AP89) Statistics Total documents 84,678 Total word occurrences 39,749,179 Vocabulary size 198,763 Words occurring > 1000 times 4,169 Words occurring once 70,064 Word Freq. r Pr(%) r.Pr Assistant 5,095 1,021 .013 0.13 Sewers 100 17,110 2.56 × 10−4 0.04 Toothbrush 10 51,555 2.56 × 10−5 0.01 Hazmat 1 166,945 2.56 × 10−6 0.04
In association with and Top 50 Words from AP89
In association with and Zipf’s Law for AP89 Note problems at high and low frequencies (Log Scale) P r (Log Scale)
In association with and Zipf’s Law ● What is the proportion of words with a given frequency? ○ Word that occurs n times has rank rn = k/n ○ Number of words with frequency n is (estimate) ■ rn − rn+1 = k/n − k/(n + 1) = k/n(n + 1) ○ Proportion found by dividing by total number of words = highest rank = k ○ So, proportion with frequency n is 1/n(n+1) ○ Proportion of words with freq n = number of distinct words with freq n / total number of unique words = k/n(n+1) / k = 1/n(n+1)
In association with and Zipf’s Law ● Example word frequency ranking ● To compute number of words with frequency 5,099 ○ rank of “chemical” minus the rank of “summit” ○ 1006 − 1002 = 4
In association with and Example ● Proportions of words occurring n times in 336,310 TREC documents ● Vocabulary size is 508,209
In association with and Vocabulary Growth ● As corpus grows, so does vocabulary size ○ Fewer new words when corpus is already large ● Observed relationship (Heaps’ Law) v = k.nβ ○ where ■ v is vocabulary size (number of unique words), ■ n is the number of words in corpus (number of total words), ■ k, β are parameters that vary for each corpus ■ (typical values given are 10 ≤ k ≤ 100 and β ≈ 0.5)
In association with and AP89 Example
In association with and Heaps’ Law Predictions ● Predictions for TREC collections are accurate for large numbers of words ○ e.g., first 10,879,522 words of the AP89 collection scanned ○ prediction is 100,151 unique words ○ actual number is 100,024 ● Predictions for small numbers of words (i.e. < 1000) are much worse
In association with and GOV2 (Web) Example
In association with and Web Example ● Heaps’ Law works with very large corpora ○ new words occurring even after seeing 30 million! ○ parameter values different than typical TREC values ● New words come from a variety of sources ■ spelling errors, invented words (e.g. product, company names), code, other languages, email addresses, etc. ● Search engines must deal with these large and growing vocabularies
In association with and Estimating Result Set Size ● How many pages contain all of the query terms? ● For the query [a b c]: ○ fabc = N · fa /N · fb /N · fc /N = (fa · fb · fc )/N2 ■ Assuming that terms occur independently ■ fabc is the estimated size of the result set ■ fa , fb , fc are the number of documents that terms a, b, and c occur in ■ N is the number of documents in the collection
In association with and GOV2 Example Collection size (N) is 25,205,179 Observed
In association with and Result Set Size Estimation ● Poor estimates because words are not independent ● Better estimates possible if co-occurrence information available ○ P(a ∩ b ∩ c) = P(a ∩ b) · P(c|(a ∩ b)) ○ ftropical∩fish∩aquarium = ftropical∩aquarium · ffish∩aquarium /faquarium ■ = 1921 · 9722/26480 = 705 ○ ftropical∩fish∩breeding = ftropical∩breeding · ffish∩breeeding /fbreeding ■ = 5510 · 36427/81885 = 2451
In association with and Result Set Estimation ● Even better estimates using initial result set ○ Estimate is simply C/s ■ where s is the proportion of the total documents that have been ranked, and C is the number of documents found that contain all the query words ○ E.g., “tropical fish aquarium” in GOV2 ■ after processing 3,000 out of the 26,480 documents that contain “aquarium”, C = 258 ● ftropical∩fish∩aquarium = 258/(3000÷26480) = 2,277 ■ After processing 20% of the documents, ● ftropical∩fish∩aquarium = 1,778 (1,529 is real value)
In association with and Estimating Collection Size ● Important issue for Web search engines ● Simple technique: use independence model ○ Given two words a and b that are independent ○ fab /N = fa /N · fb /N ○ N = (fa · fb )/fab ○ e.g., for GOV2 ■ flincoln = 771,326 ftropical = 120,990 flincoln ∩ tropical = 3,018 ■ N = (120990 · 771326)/3018 = 30,922,045 ■ (actual number is 25,205,179)
In association with and See you next time Image Credit: Adobe Text to Image
In association with and Tokenizing ● Forming words from sequence of characters ● Surprisingly complex in English, can be harder in other languages ● Early IR systems: ○ any sequence of alphanumeric characters of length 3 or more ○ terminated by a space or other special character ○ upper-case changed to lower-case
In association with and Tokenizing ● Example: ○ “Bigcorp's 2007 bi-annual report showed profits rose 10%.” becomes ○ “bigcorp 2007 annual report showed profits rose” ● Too simple for search applications or even large-scale experiments ● Why? Too much information lost ○ Small decisions in tokenizing can have major impact on effectiveness of some queries
In association with and Tokenizing Problems ● Small words can be important in some queries, usually in combinations ■ xp, ma, pm, ben e king, el paso, master p, gm, j lo, world war II ● Both hyphenated and non-hyphenated forms of many words are common ○ Sometimes hyphen is not needed ■ e-bay, wal-mart, active-x, cd-rom, t-shirts ○ At other times, hyphens should be considered either as part of the word or a word separator ■ winston-salem, mazda rx-7, e-cards, pre-diabetes, t-mobile, spanish-speaking
In association with and Tokenizing Problems ● Special characters are an important part of tags, URLs, code in documents ● Capitalized words can have different meaning from lower case words ○ Bush, Apple ● Apostrophes can be a part of a word, a part of a possessive, or just a mistake ○ rosie o'donnell, can't, don't, 80's, 1890's, men's straw hats, master's degree, england's ten largest cities, shriner's
In association with and Tokenizing Problems ● Numbers can be important, including decimals ○ nokia 3250, top 10 courses, united 93, quicktime 6.5 pro, 92.3 the beat, 288358 ● Periods can occur in numbers, abbreviations, URLs, ends of sentences, and other situations ○ I.B.M., Ph.D., cs.umass.edu, F.E.A.R. ● Note: tokenizing steps for queries must be identical to steps for documents
In association with and Tokenizing Process ● First step is to use parser to identify appropriate parts of document to tokenize ● Defer complex decisions to other components ○ word is any sequence of alphanumeric characters, terminated by a space or special character, with everything converted to lower-case ○ everything indexed ○ example: 92.3 → 92 3 but search finds documents with 92 and 3 adjacent ○ incorporate some rules to reduce dependence on query transformation components
In association with and Tokenizing Process ● Not that different than simple tokenizing process used in past ● Examples of rules used with TREC ○ Apostrophes in words ignored ■ o’connor → oconnor bob’s → bobs ○ Periods in abbreviations ignored ■ I.B.M. → ibm Ph.D. → ph d
In association with and Stopping ● Function words (determiners, prepositions) have little meaning on their own ● High occurrence frequencies ● Treated as stopwords (i.e. removed) ○ reduce index space, improve response time, improve effectiveness ● Can be important in combinations ○ e.g., “to be or not to be”
In association with and Stopping ● Stopword list can be created from high-frequency words or based on a standard list ● Lists are customized for applications, domains, and even parts of documents ○ e.g., “click” is a good stopword for anchor text ● Best policy is to index all words in documents, make decisions about which words to use at query time
In association with and Stemming ● Many morphological variations of words ○ inflectional (plurals, tenses) ○ derivational (making verbs nouns etc.) ● In most cases, these have the same or very similar meanings ● Stemmers attempt to reduce morphological variations of words to a common stem ○ usually involves removing suffixes ● Can be done at indexing time or as part of query processing (like stopwords)
In association with and Stemming ● Generally a small but significant effectiveness improvement ○ can be crucial for some languages ○ e.g., 5-10% improvement for English, up to 50% in Arabic Words with the Arabic root ktb
In association with and Stemming ● Two basic types ○ Dictionary-based: uses lists of related words ○ Algorithmic: uses program to determine related words ● Algorithmic stemmers ○ suffix-s: remove ‘s’ endings assuming plural ■ e.g., cats → cat, lakes → lake, wiis → wii ■ Many false negatives: supplies → supplie ■ Some false positives: ups → up
In association with and Porter Stemmer ● Algorithmic stemmer used in IR experiments since the 70s ● Consists of a series of rules designed to the longest possible suffix at each step ● Effective in TREC ● Produces stems not words ● Makes a number of errors and difficult to modify vvi
In association with and Porter Stemmer ● Example step (1 of 5)
In association with and Porter Stemmer ● Porter2 stemmer addresses some of these issues ● Approach has been used with other languages
In association with and Krovetz Stemmer ● Hybrid algorithmic-dictionary ○ Word checked in dictionary ■ If present, either left alone or replaced with “exception” ■ If not present, word is checked for suffixes that could be removed ■ After removal, dictionary is checked again ● Produces words not stems ● Comparable effectiveness ● Lower false positive rate, somewhat higher false negative
In association with and Stemmer Comparison
In association with and Phrases ● Many queries are 2-3 word phrases ● Phrases are ○ More precise than single words ■ e.g., documents containing “computer science” vs. two words “science” and “computer” ○ Less ambiguous ■ e.g., “big apple” vs. “apple” ● Can be difficult for ranking ■ e.g., Given query “fishing supplies”, how do we score documents with ● exact phrase many times, exact phrase just once, individual words in same sentence, same paragraph, whole document, variations on words?
In association with and Phrases ● Text processing issue – how are phrases recognized? ● Three possible approaches: ○ Identify syntactic phrases using a part-of-speech (POS) tagger ○ Use word n-grams ○ Store word positions in indexes and use proximity operators in queries
In association with and POS Tagging ● POS taggers use statistical models of text to predict syntactic tags of words ○ Example tags: ■ NN (singular noun), NNS (plural noun), VB (verb), VBD (verb, past tense), VBN (verb, past participle), IN (preposition), JJ (adjective), CC (conjunction, e.g., “and”, “or”), PRP (pronoun), and MD (modal auxiliary, e.g., “can”, “will”). ● Phrases can then be defined as simple noun groups, for example
In association with and Pos Tagging Example
In association with and Example Noun Phrases
In association with and Word N-Grams ● POS tagging too slow for large collections ● Simpler definition – phrase is any sequence of n words – known as n-grams ○ bigram: 2 word sequence, trigram: 3 word sequence, unigram: single words ○ N-grams also used at character level for applications such as OCR ● N-grams typically formed from overlapping sequences of words ○ i.e. move n-word “window” one word at a time in document
In association with and N-Grams ● Frequent n-grams are more likely to be meaningful phrases ● N-grams form a Zipf distribution ○ Better fit than words alone ● Could index all n-grams up to specified length ○ Much faster than POS tagging ○ Uses a lot of storage ■ e.g., document containing 1,000 words would contain 3,990 instances of word n-grams of length 2 ≤ n ≤ 5
In association with and Document Structure and Markup ● Some parts of documents are more important than others ● Document parser recognizes structure using markup, such as HTML tags ○ Headers, anchor text, bolded text all likely to be important ○ Metadata can also be important ○ Links used for link analysis
In association with and Example Web Page
In association with and Example Web Page
In association with and See you next time Image Credit: Adobe Text to Image
In association with and Link Analysis ● Links are a key component of the Web ● Important for navigation, but also for search ○ e.g., <a href="http://example.com" >Example website</a> ○ “Example website” is the anchor text ○ “http://example.com” is the destination link ○ both are used by search engines
In association with and Anchor Text ● Used as a description of the content of the destination page ○ i.e., collection of anchor text in all links pointing to a page used as an additional text field ● Anchor text tends to be short, descriptive, and similar to query text ● Retrieval experiments have shown that anchor text has significant impact on effectiveness for some types of queries ○ i.e., more than PageRank ○ anchor text indicates topic of the page, PageRank is topic-independent popularity of a page
In association with and Anchor Text ● Can lead to Google Bombing: anyone can call you something bad
In association with and PageRank ● Billions of web pages, some more informative than others ● Links can be viewed as information about the popularity (authority?) of a web page ○ can be used by ranking algorithm ● Inlink count could be used as simple measure ● Link analysis algorithms like PageRank provide more reliable ratings ○ less susceptible to link spam
In association with and Random Surfer Model ● Browse the Web using the following algorithm: ○ Choose a random number r between 0 and 1: λ ○ With probability λ, the web surfer goes to a random page on the web (type a new URL in their browser) ○ With probability (1-λ) he/she clicks on one of the outlinks from the page ○ If r < λ: ■ Go to a random page ○ If r ≥ λ: ■ Click a link at random on the current page ○ Start again ● PageRank of a page is the probability that the “random surfer” will be looking at that page ○ links from popular pages will increase PageRank of pages they point to
In association with and Dangling Links ● Random jump prevents getting stuck on pages that ○ do not have links ○ contains only links that no longer point to other pages ○ have links forming a loop ● Links that point to the first two types of pages are called dangling links ○ may also be links to pages that have not yet been crawled
In association with and PageRank ● PageRank (PR) of page C = PR(A)/2 + PR(B)/1 ● More generally, ○ where Bu is the set of pages that point to u, and Lv is the number of outgoing links from page v (not counting duplicate links)
In association with and PageRank ● Don’t know PageRank values at start ● Assume equal values (1/3 in this case), then iterate: ○ first iteration: PR(C) = 0.33/2 + 0.33 = 0.5, PR(A) = 0.33, and PR(B) = 0.17 ○ second: PR(C) = 0.33/2 + 0.17 = 0.33, PR(A) = 0.5, PR(B) = 0.17 ○ third: PR(C) = 0.42, PR(A) = 0.33, PR(B) = 0.25 ● Converges to PR(C) = 0.4, PR(A) = 0.4, and PR(B) = 0.2
In association with and PageRank ● Taking random page jump into account, 1/3 chance of going to any page when r < λ ● PR(C) = λ/3 + (1 − λ) · (PR(A)/2 + PR(B)/1) ● More generally, ○ where N is the number of pages in the corpus (web), λ typically 0.15
In association with and
In association with and Link Quality ● Pagerank was designed for when humans thoughtfully linked to other pages ● Link quality is affected by spam and other factors ○ e.g., link farms to increase PageRank ○ trackback links in blogs can create loops ○ links from comments section of popular blogs ■ Blog services modify comment links to contain rel=nofollow attribute ■ e.g., “Come visit my <a rel=nofollow href="http://www.page.com">web page</a>.”
In association with and Trackback Links ii

More Related Content

PPTX
Info 2402 irt-chapter_4
Shahriar Rafee
 
PPTX
Lecture 7- Text Statistics and Document Parsing
Sean Golliher
 
PDF
Chapter 2: Text Operation in information stroage and retrieval
captainmactavish1996
 
PDF
Chapter 2 Text Operation.pdf
Habtamu100
 
PDF
Chapter 2 Text Operation and Term Weighting.pdf
JemalNesre1
 
PPT
IR CHAPTER_TWO Most important for students
abduwasiahmed
 
PPT
CHapter 2_text operation.ppt material for university students
jamsibro140
 
PDF
Shilpa shukla processing_text
shilpashukla01
 
Info 2402 irt-chapter_4
Shahriar Rafee
 
Lecture 7- Text Statistics and Document Parsing
Sean Golliher
 
Chapter 2: Text Operation in information stroage and retrieval
captainmactavish1996
 
Chapter 2 Text Operation.pdf
Habtamu100
 
Chapter 2 Text Operation and Term Weighting.pdf
JemalNesre1
 
IR CHAPTER_TWO Most important for students
abduwasiahmed
 
CHapter 2_text operation.ppt material for university students
jamsibro140
 
Shilpa shukla processing_text
shilpashukla01
 

Similar to information retrival and text processing (20)

PPT
IR
Girish Khanzode
 
PDF
Information storage and Retrieval-Chapter 2 Updated.pdf
fikadumeuedu
 
PPT
2_text operationinformation retrieval. ppt
HayomeTakele
 
PDF
Information retrieval concept, practice and challenge
Gan Keng Hoon
 
PPTX
Taxonomies in Search
TSoholt
 
PPT
Information retrieval chapter 2-Text Operations.ppt
SamuelKetema1
 
PPTX
Information storage and retrieval system unit two
garedew32
 
PPT
2_text operatinnjjjjkkkkkkkkkkkkgggggggggggggggggggon.ppt
habtaassefa0
 
PPTX
Retrieval approches
Mudassar Hussain
 
PPT
Information Retrieval
ShujaatZaheer3
 
PPTX
Chapter05
pittore44
 
PPTX
01 IRS-1 (1) document upload the link to
tiggu56
 
PPTX
01 IRS to upload the data according to the.pptx
tiggu56
 
PPTX
Ir 03
Mohammed Romi
 
PDF
ICDIM 06 Web IR Tutorial [Compatibility Mode].pdf
siddiquitanveer1
 
PPT
3_Indexing.ppt
MedinaBedru
 
PDF
14. Michael Oakes (UoW) Natural Language Processing for Translation
RIILP
 
PPS
Google Search Presentation
WFL Tech Trainer, Jen Farr
 
PDF
A Simple Information Retrieval Technique
idescitation
 
PDF
Concepts and Challenges of Text Retrieval for Search Engine
Gan Keng Hoon
 
IR
Girish Khanzode
 
Information storage and Retrieval-Chapter 2 Updated.pdf
fikadumeuedu
 
2_text operationinformation retrieval. ppt
HayomeTakele
 
Information retrieval concept, practice and challenge
Gan Keng Hoon
 
Taxonomies in Search
TSoholt
 
Information retrieval chapter 2-Text Operations.ppt
SamuelKetema1
 
Information storage and retrieval system unit two
garedew32
 
2_text operatinnjjjjkkkkkkkkkkkkgggggggggggggggggggon.ppt
habtaassefa0
 
Retrieval approches
Mudassar Hussain
 
Information Retrieval
ShujaatZaheer3
 
Chapter05
pittore44
 
01 IRS-1 (1) document upload the link to
tiggu56
 
01 IRS to upload the data according to the.pptx
tiggu56
 
Ir 03
Mohammed Romi
 
ICDIM 06 Web IR Tutorial [Compatibility Mode].pdf
siddiquitanveer1
 
3_Indexing.ppt
MedinaBedru
 
14. Michael Oakes (UoW) Natural Language Processing for Translation
RIILP
 
Google Search Presentation
WFL Tech Trainer, Jen Farr
 
A Simple Information Retrieval Technique
idescitation
 
Concepts and Challenges of Text Retrieval for Search Engine
Gan Keng Hoon
 
Ad

Recently uploaded (20)

PPTX
PPH.pptx obstetrics and gynecology in nursing
SrideviDevaraj5
 
PDF
VCE English Exam - Section C Student Revision Booklet
jpinnuck
 
PDF
RMMM.pdf make it easy to upload and study
sajedul2
 
PPTX
Cell Structure & Organelles in detailed.
DHANASHREESIVAKUMAR
 
PDF
TR - Agricultural Crops Production NC III.pdf
avgilmegino3
 
PDF
STATICS OF THE RIGID BODIES Hibbelers.pdf
pascualasturiaskaye4
 
PDF
Black Hat USA 2025 - Micro ICS Summit - ICS/OT Threat Landscape
Chris Sistrunk
 
PDF
ANTIBIOTICS.pptx.pdf………………… xxxxxxxxxxxxx
HakHe
 
PDF
O5-L3 Freight Transport Ops (International) V1.pdf
labyankof
 
PDF
Supply Chain Operations Speaking Notes -ICLT Program
labyankof
 
PDF
01-Introduction-to-Information-Management.pdf
PrinceIbarra
 
PDF
Physiotherapy_for_Respiratory_and_Cardiac_Problems WEBBER.pdf
DrMONISHAphysio
 
PDF
Computing-Curriculum for Schools in Ghana
abigailanane203
 
PPTX
Introduction_to_Human_Anatomy_and_Physiology_for_B.Pharm.pptx
chetansingh379583
 
PDF
3rd Neelam Sanjeevareddy Memorial Lecture.pdf
Academy of Grassroots Studies and Research of India (AGRASRI)
 
PPTX
Lesson notes of climatology university.
sgladness67
 
PPTX
IMMUNITY IMMUNITY refers to protection against infection, and the immune syst...
shilpa939953
 
PPTX
Institutional Correction lecture only . . .
limvtheghins
 
PDF
Insiders guide to clinical Medicine.pdf
AbhishekPatil315128
 
PDF
Basic Mud Logging Guide for educational purpose
wdandworks
 
PPH.pptx obstetrics and gynecology in nursing
SrideviDevaraj5
 
VCE English Exam - Section C Student Revision Booklet
jpinnuck
 
RMMM.pdf make it easy to upload and study
sajedul2
 
Cell Structure & Organelles in detailed.
DHANASHREESIVAKUMAR
 
TR - Agricultural Crops Production NC III.pdf
avgilmegino3
 
STATICS OF THE RIGID BODIES Hibbelers.pdf
pascualasturiaskaye4
 
Black Hat USA 2025 - Micro ICS Summit - ICS/OT Threat Landscape
Chris Sistrunk
 
ANTIBIOTICS.pptx.pdf………………… xxxxxxxxxxxxx
HakHe
 
O5-L3 Freight Transport Ops (International) V1.pdf
labyankof
 
Supply Chain Operations Speaking Notes -ICLT Program
labyankof
 
01-Introduction-to-Information-Management.pdf
PrinceIbarra
 
Physiotherapy_for_Respiratory_and_Cardiac_Problems WEBBER.pdf
DrMONISHAphysio
 
Computing-Curriculum for Schools in Ghana
abigailanane203
 
Introduction_to_Human_Anatomy_and_Physiology_for_B.Pharm.pptx
chetansingh379583
 
3rd Neelam Sanjeevareddy Memorial Lecture.pdf
Academy of Grassroots Studies and Research of India (AGRASRI)
 
Lesson notes of climatology university.
sgladness67
 
IMMUNITY IMMUNITY refers to protection against infection, and the immune syst...
shilpa939953
 
Institutional Correction lecture only . . .
limvtheghins
 
Insiders guide to clinical Medicine.pdf
AbhishekPatil315128
 
Basic Mud Logging Guide for educational purpose
wdandworks
 
Ad

information retrival and text processing

  • 1. In association with and Chapter 4 Processing Text Search Engines and Information Retrieval Full Credit: Croft et al. - http://www.search-engines-book.com/
  • 2. In association with and From Ch. 2: Indexing Process
  • 3. In association with and Processing Text ● Converting documents to index terms ● Why? ○ Matching the exact string of characters typed by the user is too restrictive ■ i.e., it doesn’t work very well in terms of effectiveness ■ e.g. uppercase/lowercase ○ Not all words are of equal value in a search ○ Sometimes not clear where words begin and end ■ Not even clear what a word is in some languages ● e.g., Chinese, Korean
  • 4. In association with and Text Statistics ● Huge variety of words used in text but ● Many statistical characteristics of word occurrences are predictable ○ e.g., distribution of word counts ● Retrieval models and ranking algorithms depend heavily on statistical properties of words ○ e.g., important words occur often in documents but are not high frequency in collection Retrieval models are algorithms or frameworks used in information retrieval systems to match user queries with relevant documents from a collection of documents, such as web pages, articles, or books.
  • 5. In association with and Zipf’s Law ● Distribution of word frequencies is very skewed ○ a few words occur very often, many words hardly ever occur ○ e.g., two most common words (“the”, “of”) make up about 10% of all word occurrences in text documents ● Zipf’s “law”: ○ observation that rank (r) of a word times its frequency (f) is approximately a constant (k) ■ assuming words are ranked in order of decreasing frequency ○ i.e., r.f ≈ k or r.Pr ≈ c, where Pr is probability of word occurrence and c ≈ 0.1 for English
  • 6. In association with and Zipf’s Law
  • 7. In association with and News Collection (AP89) Statistics Total documents 84,678 Total word occurrences 39,749,179 Vocabulary size 198,763 Words occurring > 1000 times 4,169 Words occurring once 70,064 Word Freq. r Pr(%) r.Pr Assistant 5,095 1,021 .013 0.13 Sewers 100 17,110 2.56 × 10−4 0.04 Toothbrush 10 51,555 2.56 × 10−5 0.01 Hazmat 1 166,945 2.56 × 10−6 0.04
  • 8. In association with and Top 50 Words from AP89
  • 9. In association with and Zipf’s Law for AP89 Note problems at high and low frequencies (Log Scale) P r (Log Scale)
  • 10. In association with and Zipf’s Law ● What is the proportion of words with a given frequency? ○ Word that occurs n times has rank rn = k/n ○ Number of words with frequency n is (estimate) ■ rn − rn+1 = k/n − k/(n + 1) = k/n(n + 1) ○ Proportion found by dividing by total number of words = highest rank = k ○ So, proportion with frequency n is 1/n(n+1) ○ Proportion of words with freq n = number of distinct words with freq n / total number of unique words = k/n(n+1) / k = 1/n(n+1)
  • 11. In association with and Zipf’s Law ● Example word frequency ranking ● To compute number of words with frequency 5,099 ○ rank of “chemical” minus the rank of “summit” ○ 1006 − 1002 = 4
  • 12. In association with and Example ● Proportions of words occurring n times in 336,310 TREC documents ● Vocabulary size is 508,209
  • 13. In association with and Vocabulary Growth ● As corpus grows, so does vocabulary size ○ Fewer new words when corpus is already large ● Observed relationship (Heaps’ Law) v = k.nβ ○ where ■ v is vocabulary size (number of unique words), ■ n is the number of words in corpus (number of total words), ■ k, β are parameters that vary for each corpus ■ (typical values given are 10 ≤ k ≤ 100 and β ≈ 0.5)
  • 14. In association with and AP89 Example
  • 15. In association with and Heaps’ Law Predictions ● Predictions for TREC collections are accurate for large numbers of words ○ e.g., first 10,879,522 words of the AP89 collection scanned ○ prediction is 100,151 unique words ○ actual number is 100,024 ● Predictions for small numbers of words (i.e. < 1000) are much worse
  • 16. In association with and GOV2 (Web) Example
  • 17. In association with and Web Example ● Heaps’ Law works with very large corpora ○ new words occurring even after seeing 30 million! ○ parameter values different than typical TREC values ● New words come from a variety of sources ■ spelling errors, invented words (e.g. product, company names), code, other languages, email addresses, etc. ● Search engines must deal with these large and growing vocabularies
  • 18. In association with and Estimating Result Set Size ● How many pages contain all of the query terms? ● For the query [a b c]: ○ fabc = N · fa /N · fb /N · fc /N = (fa · fb · fc )/N2 ■ Assuming that terms occur independently ■ fabc is the estimated size of the result set ■ fa , fb , fc are the number of documents that terms a, b, and c occur in ■ N is the number of documents in the collection
  • 19. In association with and GOV2 Example Collection size (N) is 25,205,179 Observed
  • 20. In association with and Result Set Size Estimation ● Poor estimates because words are not independent ● Better estimates possible if co-occurrence information available ○ P(a ∩ b ∩ c) = P(a ∩ b) · P(c|(a ∩ b)) ○ ftropical∩fish∩aquarium = ftropical∩aquarium · ffish∩aquarium /faquarium ■ = 1921 · 9722/26480 = 705 ○ ftropical∩fish∩breeding = ftropical∩breeding · ffish∩breeeding /fbreeding ■ = 5510 · 36427/81885 = 2451
  • 21. In association with and Result Set Estimation ● Even better estimates using initial result set ○ Estimate is simply C/s ■ where s is the proportion of the total documents that have been ranked, and C is the number of documents found that contain all the query words ○ E.g., “tropical fish aquarium” in GOV2 ■ after processing 3,000 out of the 26,480 documents that contain “aquarium”, C = 258 ● ftropical∩fish∩aquarium = 258/(3000÷26480) = 2,277 ■ After processing 20% of the documents, ● ftropical∩fish∩aquarium = 1,778 (1,529 is real value)
  • 22. In association with and Estimating Collection Size ● Important issue for Web search engines ● Simple technique: use independence model ○ Given two words a and b that are independent ○ fab /N = fa /N · fb /N ○ N = (fa · fb )/fab ○ e.g., for GOV2 ■ flincoln = 771,326 ftropical = 120,990 flincoln ∩ tropical = 3,018 ■ N = (120990 · 771326)/3018 = 30,922,045 ■ (actual number is 25,205,179)
  • 23. In association with and See you next time Image Credit: Adobe Text to Image
  • 24. In association with and Tokenizing ● Forming words from sequence of characters ● Surprisingly complex in English, can be harder in other languages ● Early IR systems: ○ any sequence of alphanumeric characters of length 3 or more ○ terminated by a space or other special character ○ upper-case changed to lower-case
  • 25. In association with and Tokenizing ● Example: ○ “Bigcorp's 2007 bi-annual report showed profits rose 10%.” becomes ○ “bigcorp 2007 annual report showed profits rose” ● Too simple for search applications or even large-scale experiments ● Why? Too much information lost ○ Small decisions in tokenizing can have major impact on effectiveness of some queries
  • 26. In association with and Tokenizing Problems ● Small words can be important in some queries, usually in combinations ■ xp, ma, pm, ben e king, el paso, master p, gm, j lo, world war II ● Both hyphenated and non-hyphenated forms of many words are common ○ Sometimes hyphen is not needed ■ e-bay, wal-mart, active-x, cd-rom, t-shirts ○ At other times, hyphens should be considered either as part of the word or a word separator ■ winston-salem, mazda rx-7, e-cards, pre-diabetes, t-mobile, spanish-speaking
  • 27. In association with and Tokenizing Problems ● Special characters are an important part of tags, URLs, code in documents ● Capitalized words can have different meaning from lower case words ○ Bush, Apple ● Apostrophes can be a part of a word, a part of a possessive, or just a mistake ○ rosie o'donnell, can't, don't, 80's, 1890's, men's straw hats, master's degree, england's ten largest cities, shriner's
  • 28. In association with and Tokenizing Problems ● Numbers can be important, including decimals ○ nokia 3250, top 10 courses, united 93, quicktime 6.5 pro, 92.3 the beat, 288358 ● Periods can occur in numbers, abbreviations, URLs, ends of sentences, and other situations ○ I.B.M., Ph.D., cs.umass.edu, F.E.A.R. ● Note: tokenizing steps for queries must be identical to steps for documents
  • 29. In association with and Tokenizing Process ● First step is to use parser to identify appropriate parts of document to tokenize ● Defer complex decisions to other components ○ word is any sequence of alphanumeric characters, terminated by a space or special character, with everything converted to lower-case ○ everything indexed ○ example: 92.3 → 92 3 but search finds documents with 92 and 3 adjacent ○ incorporate some rules to reduce dependence on query transformation components
  • 30. In association with and Tokenizing Process ● Not that different than simple tokenizing process used in past ● Examples of rules used with TREC ○ Apostrophes in words ignored ■ o’connor → oconnor bob’s → bobs ○ Periods in abbreviations ignored ■ I.B.M. → ibm Ph.D. → ph d
  • 31. In association with and Stopping ● Function words (determiners, prepositions) have little meaning on their own ● High occurrence frequencies ● Treated as stopwords (i.e. removed) ○ reduce index space, improve response time, improve effectiveness ● Can be important in combinations ○ e.g., “to be or not to be”
  • 32. In association with and Stopping ● Stopword list can be created from high-frequency words or based on a standard list ● Lists are customized for applications, domains, and even parts of documents ○ e.g., “click” is a good stopword for anchor text ● Best policy is to index all words in documents, make decisions about which words to use at query time
  • 33. In association with and Stemming ● Many morphological variations of words ○ inflectional (plurals, tenses) ○ derivational (making verbs nouns etc.) ● In most cases, these have the same or very similar meanings ● Stemmers attempt to reduce morphological variations of words to a common stem ○ usually involves removing suffixes ● Can be done at indexing time or as part of query processing (like stopwords)
  • 34. In association with and Stemming ● Generally a small but significant effectiveness improvement ○ can be crucial for some languages ○ e.g., 5-10% improvement for English, up to 50% in Arabic Words with the Arabic root ktb
  • 35. In association with and Stemming ● Two basic types ○ Dictionary-based: uses lists of related words ○ Algorithmic: uses program to determine related words ● Algorithmic stemmers ○ suffix-s: remove ‘s’ endings assuming plural ■ e.g., cats → cat, lakes → lake, wiis → wii ■ Many false negatives: supplies → supplie ■ Some false positives: ups → up
  • 36. In association with and Porter Stemmer ● Algorithmic stemmer used in IR experiments since the 70s ● Consists of a series of rules designed to the longest possible suffix at each step ● Effective in TREC ● Produces stems not words ● Makes a number of errors and difficult to modify vvi
  • 37. In association with and Porter Stemmer ● Example step (1 of 5)
  • 38. In association with and Porter Stemmer ● Porter2 stemmer addresses some of these issues ● Approach has been used with other languages
  • 39. In association with and Krovetz Stemmer ● Hybrid algorithmic-dictionary ○ Word checked in dictionary ■ If present, either left alone or replaced with “exception” ■ If not present, word is checked for suffixes that could be removed ■ After removal, dictionary is checked again ● Produces words not stems ● Comparable effectiveness ● Lower false positive rate, somewhat higher false negative
  • 40. In association with and Stemmer Comparison
  • 41. In association with and Phrases ● Many queries are 2-3 word phrases ● Phrases are ○ More precise than single words ■ e.g., documents containing “computer science” vs. two words “science” and “computer” ○ Less ambiguous ■ e.g., “big apple” vs. “apple” ● Can be difficult for ranking ■ e.g., Given query “fishing supplies”, how do we score documents with ● exact phrase many times, exact phrase just once, individual words in same sentence, same paragraph, whole document, variations on words?
  • 42. In association with and Phrases ● Text processing issue – how are phrases recognized? ● Three possible approaches: ○ Identify syntactic phrases using a part-of-speech (POS) tagger ○ Use word n-grams ○ Store word positions in indexes and use proximity operators in queries
  • 43. In association with and POS Tagging ● POS taggers use statistical models of text to predict syntactic tags of words ○ Example tags: ■ NN (singular noun), NNS (plural noun), VB (verb), VBD (verb, past tense), VBN (verb, past participle), IN (preposition), JJ (adjective), CC (conjunction, e.g., “and”, “or”), PRP (pronoun), and MD (modal auxiliary, e.g., “can”, “will”). ● Phrases can then be defined as simple noun groups, for example
  • 44. In association with and Pos Tagging Example
  • 45. In association with and Example Noun Phrases
  • 46. In association with and Word N-Grams ● POS tagging too slow for large collections ● Simpler definition – phrase is any sequence of n words – known as n-grams ○ bigram: 2 word sequence, trigram: 3 word sequence, unigram: single words ○ N-grams also used at character level for applications such as OCR ● N-grams typically formed from overlapping sequences of words ○ i.e. move n-word “window” one word at a time in document
  • 47. In association with and N-Grams ● Frequent n-grams are more likely to be meaningful phrases ● N-grams form a Zipf distribution ○ Better fit than words alone ● Could index all n-grams up to specified length ○ Much faster than POS tagging ○ Uses a lot of storage ■ e.g., document containing 1,000 words would contain 3,990 instances of word n-grams of length 2 ≤ n ≤ 5
  • 48. In association with and Document Structure and Markup ● Some parts of documents are more important than others ● Document parser recognizes structure using markup, such as HTML tags ○ Headers, anchor text, bolded text all likely to be important ○ Metadata can also be important ○ Links used for link analysis
  • 49. In association with and Example Web Page
  • 50. In association with and Example Web Page
  • 51. In association with and See you next time Image Credit: Adobe Text to Image
  • 52. In association with and Link Analysis ● Links are a key component of the Web ● Important for navigation, but also for search ○ e.g., <a href="http://example.com" >Example website</a> ○ “Example website” is the anchor text ○ “http://example.com” is the destination link ○ both are used by search engines
  • 53. In association with and Anchor Text ● Used as a description of the content of the destination page ○ i.e., collection of anchor text in all links pointing to a page used as an additional text field ● Anchor text tends to be short, descriptive, and similar to query text ● Retrieval experiments have shown that anchor text has significant impact on effectiveness for some types of queries ○ i.e., more than PageRank ○ anchor text indicates topic of the page, PageRank is topic-independent popularity of a page
  • 54. In association with and Anchor Text ● Can lead to Google Bombing: anyone can call you something bad
  • 55. In association with and PageRank ● Billions of web pages, some more informative than others ● Links can be viewed as information about the popularity (authority?) of a web page ○ can be used by ranking algorithm ● Inlink count could be used as simple measure ● Link analysis algorithms like PageRank provide more reliable ratings ○ less susceptible to link spam
  • 56. In association with and Random Surfer Model ● Browse the Web using the following algorithm: ○ Choose a random number r between 0 and 1: λ ○ With probability λ, the web surfer goes to a random page on the web (type a new URL in their browser) ○ With probability (1-λ) he/she clicks on one of the outlinks from the page ○ If r < λ: ■ Go to a random page ○ If r ≥ λ: ■ Click a link at random on the current page ○ Start again ● PageRank of a page is the probability that the “random surfer” will be looking at that page ○ links from popular pages will increase PageRank of pages they point to
  • 57. In association with and Dangling Links ● Random jump prevents getting stuck on pages that ○ do not have links ○ contains only links that no longer point to other pages ○ have links forming a loop ● Links that point to the first two types of pages are called dangling links ○ may also be links to pages that have not yet been crawled
  • 58. In association with and PageRank ● PageRank (PR) of page C = PR(A)/2 + PR(B)/1 ● More generally, ○ where Bu is the set of pages that point to u, and Lv is the number of outgoing links from page v (not counting duplicate links)
  • 59. In association with and PageRank ● Don’t know PageRank values at start ● Assume equal values (1/3 in this case), then iterate: ○ first iteration: PR(C) = 0.33/2 + 0.33 = 0.5, PR(A) = 0.33, and PR(B) = 0.17 ○ second: PR(C) = 0.33/2 + 0.17 = 0.33, PR(A) = 0.5, PR(B) = 0.17 ○ third: PR(C) = 0.42, PR(A) = 0.33, PR(B) = 0.25 ● Converges to PR(C) = 0.4, PR(A) = 0.4, and PR(B) = 0.2
  • 60. In association with and PageRank ● Taking random page jump into account, 1/3 chance of going to any page when r < λ ● PR(C) = λ/3 + (1 − λ) · (PR(A)/2 + PR(B)/1) ● More generally, ○ where N is the number of pages in the corpus (web), λ typically 0.15
  • 62. In association with and Link Quality ● Pagerank was designed for when humans thoughtfully linked to other pages ● Link quality is affected by spam and other factors ○ e.g., link farms to increase PageRank ○ trackback links in blogs can create loops ○ links from comments section of popular blogs ■ Blog services modify comment links to contain rel=nofollow attribute ■ e.g., “Come visit my <a rel=nofollow href="http://www.page.com">web page</a>.”
  • 63. In association with and Trackback Links ii