FullText Indexing Session 10 INFM 718 N WebEnabled

Full-Text Indexing Session 10 INFM 718 N Web-Enabled Databases

Agenda • How to do it • How it works • The “A” Team

• Relational normalization • Structured programming • Software patterns • Object-oriented design • Functional decomposition Business Interaction Design rules Interface Design Client Hardware Web Browser Client-side Programming Interchange Language Server-side Programming (PC) (IE, Firefox) (Java. Script) (HTML, XML) (PHP) Database (My. SQL) Server Hardware (PC, Unix)

Full-Text Indexing in My. SQL • Create a My. ISAM table (not Inno. DB!) – Include a CHAR, VARCHAR, or TEXT field – Text fields can hold a bit over 10, 000 words • Create a FULLTEXT index – ALTER TABLE x ADD FULLTEXT INDEX y; • Issue a (ranked) query – SELECT y FROM x WHERE MATCH y AGAINST (‘cat’);

Other Types of Queries • Automatic (ranked) vocabulary expansion – SELECT y FROM x WHERE MATCH y AGAINST (‘cat’ WITH QUERY EXPANSION); • Boolean (unranked) search – SELECT y FROM x WHERE MATCH y AGAINST (‘+cat -dog’ IN BOOLEAN MODE);

Query Details • No more than 254 characters (~40 words) – Longer queries take more time • Multiple words are implicitly joined by “OR” • Boolean queries can use (unnested) operators – Words preceded by “+” must occur (AND) – Words preceded by “-” must not occur (AND NOT)

What’s a “Word? ” • Delimited by “white space” or “-” – White-space includes space, tab, newline, … • Not case sensitive • Exact string match – No “stemming” (automatic truncation) • Boolean search has additional options – Truncation (e. g. , time*) – Phrases (e. g. , “cats and dogs”)

Unsearchable Words • Very common words – Those that appear in more than 50% of docs • Words of 3 or fewer characters – Rarely are topically specific • Other “stopwords” – able about above accordingly across actually afterwards against ain't …

Human-Machine Synergy • Machines are good at: – Doing simple things accurately and quickly – Scaling to larger collections in sublinear time • People are better at: – Accurately recognizing what they are looking for – Evaluating intangibles such as “quality” • Both are pretty bad at: – Mapping consistently between words and concepts

Supporting the Search Process Source Selection IR System Query Formulation Predict Nominate Choose Query Search Query Reformulation and Relevance Feedback Ranked List Selection Document Examination Source Reselection Document Delivery

Supporting the Search Process Source Selection IR System Query Formulation Query Search Ranked List Selection Indexing Document Index Examination Acquisition Document Collection Delivery

Q 1 Visceral Need Q 2 Conscious Need Q 3 Formalized Need Q 4 Compromised Need (Query) Intermediated Search End-user Search Taylor’s Model of Question Formation

Search Goal • Choose the same documents a human would – Without human intervention (less work) – Faster than a human could (less time) – As accurately as possible (less accuracy) • Humans start with an information need – Machines start with a query • Humans match documents to information needs – Machines match document & query representations

Search Component Model Utility Human Judgment Document Query Formulation Query Representation Function Query Representation Document Representation Comparison Function Retrieval Status Value Document Processing Query Processing Information Need

Relevance • Relevance relates a topic and a document – Duplicates are equally relevant, by definition – Constant over time and across users • Pertinence relates a task and a document – Accounts for quality, complexity, language, … • Utility relates a user and a document – Accounts for prior knowledge • We seek utility, but relevance is what we get!

Problems With Word Matching • Word matching suffers from two problems – Synonymy: paper vs. article – Homonymy: bank (river) vs. bank (financial) • Disambiguation in IR: seek to resolve homonymy – Index word senses rather than words • Synonymy usually addressed by – Thesaurus-based query expansion – Latent semantic indexing

“Bag of Terms” Representation • Bag = a “set” that can contain duplicates Ø “The quick brown fox jumped over the lazy dog’s back” {back, brown, dog, fox, jump, lazy, over, quick, the} • Vector = values recorded in any consistent order Ø {back, brown, dog, fox, jump, lazy, over, quick, the} [1 1 1 1 2]

Document 1 The quick brown fox jumped over the lazy dog’s back. Document 2 Now is the time for all good men to come to the aid of their party. Term aid all back brown come dog fox good jump lazy men now over party quick their time Document 1 Document 2 Bag of Terms Example 0 0 1 1 0 0 1 0 0 1 1 0 1 1 Stopword List for is of the to

Boolean IR • Strong points – Accurate, if you know the right strategies – Efficient for the computer • Weaknesses – Often results in too many documents, or none – Users must learn Boolean logic – Sometimes finds relationships that don’t exist – Words can have many meanings – Choosing the right words is sometimes hard

Proximity Operators • More precise versions of AND – “NEAR n” allows at most n-1 intervening terms – “WITH” requires terms to be adjacent and in order • Easy to implement, but less efficient – Store a list of positions for each word in each doc • Stopwords become very important! – Perform normal Boolean computations • Treat WITH and NEAR like AND with an extra constraint

aid all back brown come dog fox good jump lazy men now over party quick their time Doc 2 Term Doc 1 Proximity Operator Example 0 1 (13) 0 1 (6) 1 (10) 0 1 (3) 0 0 1 (9) 0 1 (4) 0 0 1 (7) 1 (5) 0 1 (8) 0 1 (1) 1 (6) 0 0 1 (16) 1 (2) 0 0 1 (15) 0 1 (4) • time AND come – Doc 2 • time (NEAR 2) come – Empty • quick (NEAR 2) fox – Doc 1 • quick WITH fox – Empty

Advantages of Ranked Retrieval • Closer to the way people think – Some documents are better than others • Enriches browsing behavior – Decide how far down the list to go as you read it • Allows more flexible queries – Long and short queries can produce useful results

Ranked Retrieval Challenges • “Best first” is easy to say but hard to do! – The best we can hope for is to approximate it • Will the user understand the process? – It is hard to use a tool that you don’t understand • Efficiency becomes a concern – Only a problem for long queries, though

Similarity-Based Queries • Treat the query as if it were a document – Create a query bag-of-words • Find the similarity of each document – Using the coordination measure, for example • Rank order the documents by similarity – Most similar to the query first • Surprisingly, this works pretty well! – Especially for very short queries

Counting Terms • Terms tell us about documents – If “rabbit” appears a lot, it may be about rabbits • Documents tell us about terms – “the” is in every document -- not discriminating • Documents are most likely described well by rare terms that occur in them frequently – Higher “term frequency” is stronger evidence – Low “collection frequency” makes it stronger still

The Document Length Effect • Humans look for documents with useful parts – But probabilities are computed for the whole • Document lengths vary in many collections – So probability calculations could be inconsistent • Two strategies – Adjust probability estimates for document length – Divide the documents into equal “passages”

Incorporating Term Frequency • High term frequency is evidence of meaning – And high IDF is evidence of term importance • Recompute the bag-of-words – Compute TF * IDF for every element

TF*IDF Example 1 2 complicated contaminated fallout information 4 5 6 interesting nuclear 3 1 3 6 retrieval siberia 1 2 3 4 5 3 4 3 2 7 1 3 2 4 1 2 0. 301 0. 125 0. 000 0. 602 0. 301 0. 125 0. 602 3 4 1. 51 0. 60 0. 50 0. 13 0. 38 0. 63 0. 50 0. 38 0. 60 0. 90 Weighted query: contaminated(3) retrieval(1) Result: 1, 3, 2, 4 2. 11 0. 75 0. 13 0. 50 1. 20 Unweighted query: contaminated retrieval Result: 2, 3, 1, 4 IDF-weighted query: contaminated retrieval Result: 2, 3, 1, 4

Document Length Normalization • Long documents have an unfair advantage – They use a lot of terms • So they get more matches than short documents – And they use the same words repeatedly • So they have much higher term frequencies • Normalization seeks to remove these effects – Related somehow to maximum term frequency – But also sensitive to the of number of terms

“Okapi” Term Weights TF component IDF component

My. SQL Term Weights • local weight = (log(tf)+1)/sumtf * U/(1+0. 0115*U) • global weight = log((N-nf)/nf) • query weight = local weight * global weight * qf tf sumtf U N nf qf How many times the term appears in the row The sum of "(log(tf)+1)" for all terms in the same row How many unique terms are in the row How many rows are in the table How many rows contain the term How many times the term appears in the query

Summary • Goal: find documents most similar to the query • Compute normalized document term weights – Some combination of TF, DF, and Length • Optionally, get query term weights from the user – Estimate of term importance • Compute inner product of query and doc vectors – Multiply corresponding elements and then add

Inverted File A B AI AL BA BR C D F G J L M N O P Q T TH TI Term aid all back brown come dog fox good jump lazy men now over party quick their time Doc 1 Doc 2 Doc 3 Doc 4 Doc 5 Doc 6 Doc 7 Doc 8 The Indexing Process 0 0 1 1 0 0 0 1 0 1 1 0 0 1 0 0 1 1 0 0 1 1 0 0 0 0 0 1 0 1 1 0 0 1 0 0 0 1 0 0 1 0 0 1 0 0 1 1 0 0 0 Postings 4, 8 2, 4, 6 1, 3, 7 1, 3, 5, 7 2, 4, 6, 8 3 1, 3, 5, 7 2, 4, 8 2, 6, 8 1, 3, 5, 7, 8 6, 8 1, 3 1, 5, 7 2, 4, 6

The Finished Product Inverted File A B AI AL BA BR C D F G J L M N O P Q T TH TI Term Postings aid all back brown come dog fox good jump lazy men now over party quick their time 4, 8 2, 4, 6 1, 3, 7 1, 3, 5, 7 2, 4, 6, 8 3 1, 3, 5, 7 2, 4, 8 2, 6, 8 1, 3, 5, 7, 8 6, 8 1, 3 1, 5, 7 2, 4, 6

How Big Is the Postings File? • Very compact for Boolean retrieval – About 10% of the size of the documents • If an aggressive stopword list is used! • Not much larger for ranked retrieval – Perhaps 20% • Enormous for proximity operators – Sometimes larger than the documents!

Building an Inverted Index • Simplest solution is a single sorted array – Fast lookup using binary search – But sorting large files on disk is very slow – And adding one document means starting over • Tree structures allow easy insertion – But the worst case lookup time is linear • Balanced trees provide the best of both – Fast lookup and easy insertion – But they require 45% more disk space

How Big is the Inverted Index? • Typically smaller than the postings file – Depends on number of terms, not documents • Eventually, most terms will already be indexed – But the postings file will continue to grow • Postings dominate asymptotic space complexity – Linear in the number of documents

Summary • Slow indexing yields fast query processing – Key fact: most terms don’t appear in most documents • We use extra disk space to save query time – Index space is in addition to document space – Time and space complexity must be balanced • Disk block reads are the critical resource – This makes index compression a big win