PrasadL09VSM-Ranking1 Vector Space Model : Ranking Revisited Adapted from Lectures by Prabhakar Raghavan (Yahoo and Stanford) and Christopher Manning (Stanford)

Slides:



Advertisements
Similar presentations
Introduction to Information Retrieval Introduction to Information Retrieval Lecture 7: Scoring and results assembly.
Advertisements

Introduction to Information Retrieval Introduction to Information Retrieval CS276 Information Retrieval and Web Search Chris Manning, Pandu Nayak and Prabhakar.
Faster TF-IDF David Kauchak cs160 Fall 2009 adapted from:
Information Retrieval CSE 8337 (Part III) Spring 2011 Some Material for these slides obtained from: Modern Information Retrieval by Ricardo Baeza-Yates.
Srihari-CSE535-Spring2008 CSE 535 Information Retrieval Lecture 2: Boolean Retrieval Model.
Information Retrieval Lecture 6bis. Recap: tf-idf weighting The tf-idf weight of a term is the product of its tf weight and its idf weight. Best known.
| 1 › Gertjan van Noord2014 Zoekmachines Lecture 4.
Ranking models in IR Key idea: We wish to return in order the documents most likely to be useful to the searcher To do this, we want to know which documents.
Introduction to Information Retrieval (Manning, Raghavan, Schutze) Chapter 6 Scoring term weighting and the vector space model.
TF-IDF David Kauchak cs160 Fall 2009 adapted from:
From last time What’s the real point of using vector spaces?: A user’s query can be viewed as a (very) short document. Query becomes a vector in the same.
CS347 Lecture 4 April 18, 2001 ©Prabhakar Raghavan.
Parametric search and zone weighting Lecture 6. Recap of lecture 4 Query expansion Index construction.
CS276 Information Retrieval and Web Mining
Hinrich Schütze and Christina Lioma
Introduction to Information Retrieval Introduction to Information Retrieval Hinrich Schütze and Christina Lioma Lecture 7: Scores in a Complete Search.
Algoritmi per IR Ranking. The big fight: find the best ranking...
Information Retrieval IR 6. Recap of the last lecture Parametric and field searches Zones in documents Scoring documents: zone weighting Index support.
CpSc 881: Information Retrieval. 2 Why is ranking so important? Problems with unranked retrieval Users want to look at a few results – not thousands.
Vector Space Model : TF - IDF
CES 514 Data Mining March 11, 2010 Lecture 5: scoring, term weighting, vector space model (Ch 6)
Chapter 5: Information Retrieval and Web Search
CSCI 5417 Information Retrieval Systems Jim Martin Lecture 6 9/8/2011.
CS276A Information Retrieval Lecture 7. Recap of the last lecture Parametric and field searches Zones in documents Scoring documents: zone weighting Index.
Information Retrieval Basic Document Scoring. Similarity between binary vectors Document is binary vector X,Y in {0,1} v Score: overlap measure What’s.
Web search basics (Recap) The Web Web crawler Indexer Search User Indexes Query Engine 1 Ad indexes.
Document ranking Paolo Ferragina Dipartimento di Informatica Università di Pisa.
Information Retrieval Lecture 6. Recap of the last lecture Parametric and field searches Zones in documents Scoring documents: zone weighting Index support.
Introduction to Information Retrieval Scores in a Complete Search System CSE 538 MRS BOOK – CHAPTER VII 1.
Faster TF-IDF David Kauchak cs458 Fall 2012 adapted from:
Hyperlink Analysis for the Web. Information Retrieval Input: Document collection Goal: Retrieve documents or text with information content that is relevant.
PrasadL09VSM-Ranking1 Vector Space Model : Ranking Revisited Adapted from Lectures by Prabhakar Raghavan (Google) and Christopher Manning (Stanford)
Computing Scores in a Complete Search System Lecture 8: Scoring and results assembly Web Search and Mining 1.
Information Retrieval and Web Search Lecture 7 1.
Term Weighting and Ranking Models Debapriyo Majumdar Information Retrieval – Spring 2015 Indian Statistical Institute Kolkata.
CSE 6331 © Leonidas Fegaras Information Retrieval 1 Information Retrieval and Web Search Engines Leonidas Fegaras.
Information Retrieval Lecture 2 Introduction to Information Retrieval (Manning et al. 2007) Chapter 6 & 7 For the MSc Computer Science Programme Dell Zhang.
Inverted index, Compressing inverted index And Computing score in complete search system Chintan Mistry Mrugank dalal.
Hyperlink Analysis for the Web. Information Retrieval Input: Document collection Goal: Retrieve documents or text with information content that is relevant.
Advanced topics in Computer Science Jiaheng Lu Department of Computer Science Renmin University of China
1 Motivation Web query is usually two or three words long. –Prone to ambiguity –Example “keyboard” –Input device of computer –Musical instruments How can.
Term Frequency. Term frequency Two factors: – A term that appears just once in a document is probably not as significant as a term that appears a number.
Information Retrieval and Data Mining (AT71. 07) Comp. Sc. and Inf
Chapter 6: Information Retrieval and Web Search
1 ITCS 6265 Information Retrieval and Web Mining Lecture 7.
Introduction to Information Retrieval (Manning, Raghavan, Schutze) Chapter 7 Computing scores in a complete search system.
Vector Space Models.
Search Engines WS 2009 / 2010 Prof. Dr. Hannah Bast Chair of Algorithms and Data Structures Department of Computer Science University of Freiburg Lecture.
Introduction to Information Retrieval Introduction to Information Retrieval Modified from Stanford CS276 slides Chap. 7: Scoring and results assembly.
Lecture 6: Scoring, Term Weighting and the Vector Space Model
Information Retrieval Techniques MS(CS) Lecture 7 AIR UNIVERSITY MULTAN CAMPUS Most of the slides adapted from IIR book.
Introduction to Information Retrieval Introduction to Information Retrieval CS276 Information Retrieval and Web Search Chris Manning and Pandu Nayak Efficient.
CS276 Information Retrieval and Web Search Lecture 7: Vector space retrieval.
Term weighting and Vector space retrieval
Introduction to Information Retrieval Introduction to Information Retrieval CS276 Information Retrieval and Web Search Christopher Manning and Prabhakar.
Introduction to Information Retrieval Introduction to Information Retrieval Lecture 9: Scoring, Term Weighting and the Vector Space Model.
Information Retrieval and Data Mining (AT71. 07) Comp. Sc. and Inf
Information Retrieval and Web Search
Query processing: phrase queries and positional indexes
Prof. Paolo Ferragina, Algoritmi per "Information Retrieval"
CSE 538 MRS BOOK – CHAPTER VII
Prof. Paolo Ferragina, Algoritmi per "Information Retrieval"
8. Efficient Scoring Most slides were adapted from Stanford CS 276 course and University of Munich IR course.
Query processing: phrase queries and positional indexes
CS276 Lecture 7: Scoring and results assembly
CS276 Information Retrieval and Web Search
Presentation transcript:

PrasadL09VSM-Ranking1 Vector Space Model : Ranking Revisited Adapted from Lectures by Prabhakar Raghavan (Yahoo and Stanford) and Christopher Manning (Stanford)

Recap: 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 Increases with the number of occurrences within a document Increases with the rarity of the term in the collection Prasad2L09VSM-Ranking

Recap: 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 Prasad3L09VSM-Ranking

Recap: cosine(query,document) Dot product Unit vectors cos(q,d) is the cosine similarity of q and d … or, equivalently, the cosine of the angle between q and d. Prasad4L09VSM-Ranking

This lecture Speeding up vector space ranking Putting together a complete search system Will require learning about a number of miscellaneous topics and heuristics Prasad5L09VSM-Ranking

Computing cosine scores

Efficient cosine ranking Find the K docs in the collection “nearest” to the query  K largest query-doc cosines. Efficient ranking: Computing a single cosine efficiently. Choosing the K largest cosine values efficiently. Can we do this without computing all N cosines? Prasad7L09VSM-Ranking

(cont’d) What we’re doing in effect: solving the K- nearest neighbor problem for a query vector In general, we do not know how to do this efficiently for high-dimensional spaces But it is solvable for short queries, and standard indexes support this well Prasad8L09VSM-Ranking

Special case – unweighted queries No weighting on query terms Assume each query term occurs only once Then for ranking, don’t need to normalize query vector Slight simplification 7.1 Prasad9L09VSM-Ranking

Faster cosine: unweighted query 7.1 Read redundant?

Computing the K largest cosines: selection vs. sorting Typically we want to retrieve the top K docs (in the cosine ranking for the query) not to totally order all docs in the collection Can we pick off docs with K highest cosines? Let J = number of docs with nonzero cosines We seek the K best of these J 7.1 Prasad11L09VSM-Ranking

Use heap for selecting top K Binary tree in which each node’s value > the values of children Takes 2J operations to construct, then each of K “winners” read off in 2log J steps. For J=10M, K=100, this is about 10% of the cost of sorting Prasad12L09VSM-Ranking

Bottlenecks Primary computational bottleneck in scoring: cosine computation Can we avoid all this computation? Yes, but may sometimes get it wrong a doc not in the top K may creep into the list of K output docs Is this such a bad thing? Prasad13L09VSM-Ranking

Cosine similarity is only a proxy User has a task and a query formulation Cosine matches docs to query Thus cosine is anyway a proxy for user happiness If we get a list of K docs “close” to the top K by cosine measure, should be ok Prasad14L09VSM-Ranking

Generic approach Find a set A of contenders, with K < |A| << N A does not necessarily contain the top K, but has many docs from among the top K Return the top K docs in A Think of A as pruning non-contenders The same approach is also used for other (non-cosine) scoring functions Will look at several schemes following this approach PrasadL09VSM-Ranking

Index elimination Basic algorithm of Fig 7.1 considers only docs containing at least one query term Take this further: Consider only high-idf query terms Consider only docs containing many query terms Prasad16L09VSM-Ranking

High-idf query terms only For a query such as catcher in the rye Only accumulate scores from catcher and rye Intuition: in and the contribute little to the scores and don’t alter rank-ordering much Benefit: Postings of low-idf terms have many docs  these (many) docs get eliminated from A Prasad17L09VSM-Ranking

Docs containing many query terms Any doc with at least one query term is a candidate for the top K output list For multi-term queries, only compute scores for docs containing several of the query terms Say, at least 3 out of 4 Imposes a “soft conjunction” on queries seen on web search engines (early Google) Easy to implement in postings traversal Prasad18L09VSM-Ranking

3 of 4 query terms Brutus Caesar Calpurnia Antony Scores only computed for 8, 16 and Prasad19L09VSM-Ranking

Champion lists Precompute for each dictionary term t, the r docs of highest weight in t’s postings Call this the champion list for t (aka fancy list or top docs for t) Note that r has to be chosen at index time At query time, only compute scores for docs in the champion list of some query term Pick the K top-scoring docs from amongst these Prasad20L09VSM-Ranking

Exercises How do Champion Lists relate to Index Elimination? Can they be used together? How can Champion Lists be implemented in an inverted index? Note the champion list has nothing to do with small docIDs Prasad21L09VSM-Ranking

Quantitative Static quality scores We want top-ranking documents to be both relevant and authoritative Relevance is being modeled by cosine scores Authority is typically a query-independent property of a document Examples of authority signals Wikipedia among websites Articles in certain newspapers A paper with many citations Many diggs, Y!buzzes or del.icio.us marks (Pagerank) 7.1.4

Modeling authority Assign to each document a query-independent quality score in [0,1] to each document d Denote this by g(d) Thus, a quantity like the number of citations is scaled into [0,1] Exercise: suggest a formula for this Prasad23L09VSM-Ranking

Net score Consider a simple total score combining cosine relevance and authority net-score(q,d) = g(d) + cosine(q,d) Can use some other linear combination than an equal weighting Indeed, any function of the two “signals” of user happiness – more later Now we seek the top K docs by net score Prasad24L09VSM-Ranking

Top K by net score – fast methods First idea: Order all postings by g(d) Key: this is a common ordering for all postings Thus, can concurrently traverse query terms’ postings for Postings intersection Cosine score computation Exercise: write pseudocode for cosine score computation if postings are ordered by g(d) Prasad25L09VSM-Ranking

Why order postings by g(d)? Under g(d)-ordering, top-scoring docs likely to appear early in postings traversal In time-bound applications (say, we have to return whatever search results we can in 50 ms), this allows us to stop postings traversal early Short of computing scores for all docs in postings Prasad26L09VSM-Ranking

Champion lists in g(d)-ordering Can combine champion lists with g(d)-ordering Maintain for each term a champion list of the r docs with highest g(d) + tf-idf td Seek top-K results from only the docs in these champion lists Prasad27L09VSM-Ranking

High and low lists For each term, we maintain two postings lists called high and low Think of high as the champion list When traversing postings on a query, only traverse high lists first If we get more than K docs, select the top K and stop Else proceed to get docs from the low lists Can be used even for simple cosine scores, without global quality g(d) A means for segmenting index into two tiers 7.1.4

Impact-ordered postings We only want to compute scores for docs for which wf t,d is high enough We sort each postings list by wf t,d Now: not all postings in a common order! How do we compute scores in order to pick off top K? Two ideas follow Prasad29L09VSM-Ranking

1. Early termination When traversing t’s postings, stop early after either a fixed number of r docs wf t,d drops below some threshold Take the union of the resulting sets of docs One from the postings of each query term Compute only the scores for docs in this union Prasad30L09VSM-Ranking

2. idf-ordered terms When considering the postings of query terms Look at them in order of decreasing idf High idf terms likely to contribute most to score As we update score contribution from each query term Stop if doc scores relatively unchanged Can apply to cosine or some other net scores Prasad31L09VSM-Ranking

Cluster pruning: preprocessing Pick  N docs at random: call these leaders For every other doc, pre-compute nearest leader Docs attached to a leader: its followers; Likely: each leader has ~  N followers Prasad32L09VSM-Ranking

Cluster pruning: query processing Process a query as follows: Given query Q, find its nearest leader L. Seek K nearest docs from among L’s followers Prasad33L09VSM-Ranking

Visualization Query LeaderFollower Prasad34L09VSM-Ranking

Why use random sampling Fast Leaders reflect data distribution Prasad35L09VSM-Ranking

General variants Have each follower attached to b1=3 (say) nearest leaders. From query, find b2=4 (say) nearest leaders and their followers. Can recur on leader/follower construction Prasad36L09VSM-Ranking

Exercises To find the nearest leader in step 1, how many cosine computations do we do? Why did we have  N in the first place? What is the effect of the constants b1, b2 on the previous slide? Devise an example where this is likely to fail – i.e., we miss one of the K nearest docs. Likely under random sampling Prasad37L09VSM-Ranking

Parametric and zone indexes Thus far, a doc has been a sequence of terms In fact documents have multiple parts, some with special semantics: Author Title Date of publication Language Format etc. These constitute the metadata about a document 6.1 Prasad38L09VSM-Ranking

Fields We sometimes wish to search by these metadata E.g., find docs authored by William Shakespeare in the year 1601, containing alas poor Yorick Year = 1601 is an example of a field Also, author last name = shakespeare, etc Field or parametric index: postings for each field value Sometimes build range trees (e.g., for dates) Field query typically treated as conjunction (doc must be authored by shakespeare) 6.1 Prasad39L09VSM-Ranking

Zone A zone is a region of the doc that can contain an arbitrary amount of text e.g., Title Abstract References … Build inverted indexes on zones as well to permit querying E.g., “find docs with merchant in the title zone and matching the query gentle rain” 6.1 Prasad40L09VSM-Ranking

Example zone indexes 6.1 Encode zones in dictionary vs. postings.

Tiered indexes Break postings up into a hierarchy of lists Most important … Least important Can be done by g(d) or another measure Inverted index thus broken up into tiers of decreasing importance At query time use top tier unless it fails to yield K docs If so drop to lower tiers Prasad42L09VSM-Ranking

Example tiered index Prasad

Query term proximity Free text queries: just a set of terms typed into the query box – common on the web Users prefer docs in which query terms occur within close proximity of each other Let w be the smallest window in a doc containing all query terms, e.g., For the query strained mercy the smallest window in the doc The quality of mercy is not strained is 4 (words) Would like scoring function to take this into account – how? L09VSM-Ranking

Query parsers Free text query from user may in fact spawn one or more queries to the indexes, e.g. query rising interest rates Run the query as a phrase query If <K docs contain the phrase rising interest rates, run the two phrase queries rising interest and interest rates If we still have <K docs, run the vector space query rising interest rates Rank matching docs by vector space scoring This sequence is issued by a query parser Prasad45L09VSM-Ranking

Aggregate scores We’ve seen that score functions can combine cosine, static quality, proximity, etc. How do we know the best combination? Some applications – expert-tuned Increasingly common: machine-learned Prasad46L09VSM-Ranking

Putting it all together Prasad47L09VSM-Ranking