Machine Learning and Data Mining 19 Mining Text and Web Data 26716

7/28/2019 Machine Learning and Data Mining 19 Mining Text and Web Data 26716

1/75

Prof. Pier Luca Lanzi

Text and Web MiningMachine Learning and Data Mining (Unit 19)


2/75

2


References

Jiawei Han and Micheline Kamber,"Data Mining: Concepts andTechniques", The Morgan KaufmannSeries in Data Management Systems(Second Edition)

Chapter 10, part 2

Web Mining Course byGregory-Platesky Shapiro available atwww.kdnuggets.com


3/75

3


Mining Text Data: An Introduction

Data Mining / Knowledge Discovery

Structured Data Multimedia Free Text Hypertext

HomeLoan (Loanee: Frank RizzoLender: MWF

Agency: Lake View Amount: $200,000

Term: 15 years)

Frank Rizzo boughthis home from LakeView Real Estate in1992.

He paid $200,000

under a15-year loanfrom MW Financial.

Frank Rizzo Bought this home from LakeView Real Estate In 1992 .

...Loans( $200K ,[map ],... )


4/75

4


Bag-of-Tokens Approaches

Four score and sevenyears ago our fathers broughtforth on this continent, a newnation , conceived in Liberty,and dedicated to theproposition that all men arecreated equal.

Now we are engaged in agreat civil war, testing

whether that nation , or

nation 5civil - 1war 2men 2

died 4people 5Liberty 1God 1

FeatureExtraction

Loses all order-specific information!Severely limits context!

Documents Token Sets


5/75

5


Natural Language Processing

A dog is chasing a boy on the playgroundDet Noun Aux Verb Det Noun Prep Det Noun

Noun Phrase Complex Verb Noun PhraseNoun Phrase

Prep PhraseVerb Phrase

Verb Phrase

Sentence

Dog(d1).Boy(b1).Playground(p1).Chasing(d1,b1,p1).

Semantic analysis

Lexicalanalysis

(part-of-speechtagging)

Syntactic analysis

(Parsing)

A person saying this maybe reminding another person to

get the dog back

Pragmatic analysis(speech act)

Scared(x) if Chasing(_,x,_).

+

Scared(b1)Inference

(Taken from ChengXiang Zhai, CS 397cxz Fall 2003)


6/75

6


General NLPToo Difficult!

Word-level ambiguity design can be a noun or a verb (Ambiguous POS) root has multiple meanings (Ambiguous sense)

Syntactic ambiguity natural language processing (Modification) A man saw a boy with a telescope. (PP Attachment)

Anaphora resolution John persuaded Bill to buy a TV for himself. (himself = John or Bill?)

Presupposition

He has quit smoking. implies that he smoked before.

(Taken from ChengXiang Zhai, CS 397cxz Fall 2003)

Humans rely on context to interpret (when possible).This context may extend beyond a given document!


7/75

7


Shallow Linguistics

English LexiconPart-of-Speech TaggingWord Sense Disambiguation

Phrase Detection / Parsing


8/75

8


WordNet

An extensive lexical network for the English languageContains over 138,838 words.Several graphs, one for each part-of-speech.Synsets (synonym sets), each defining a semantic sense.Relationship information (antonym, hyponym, meronym )Downloadable for free (UNIX, Windows)Expanding to other languages (Global WordNet Association)

Funded >$3 million, mainly government (translation interest)Founder George Miller, National Medal of Science, 1991.

wet dry

watery

moist

damp

parched

anhydrous

aridsynonym

antonym


9/75

9


1 1

1 1 1

11

( ,..., , ,..., )

( | )... ( | ) ( )... ( )

( | ) ( | )

k k

k k k

k

i i i ii

p w w t t

p t w p t w p w p w

p w t p t t =

=

1 1

1 1 1

11

( ,..., , ,..., )

( | )... ( | ) ( )... ( )

( | ) ( | )

k k

k k k

k

i i i ii

p w w t t

p t w p t w p w p w

p w t p t t =

=

Pick the most likely tag sequence.

Part-of-Speech Tagging

This sentence serves as an example of annotated textDet N V1 P Det N P V2 N

Training data (Annotated text)

POS TaggerThis is a new sentence. This is a new sentence.Det Aux Det Adj N

Partial dependency(HMM)

Independent assignmentMost common tag


10/75

10


Word Sense Disambiguation

Supervised LearningFeatures:

Neighboring POS tags (N Aux V P N)

Neighboring words (linguistics are rooted in ambiguity)Stemmed form (root)Dictionary/Thesaurus entries of neighboring wordsHigh co-occurrence words (plant, tree, origin,)Other senses of word within discourse

Algorithms:Rule-based Learning (e.g. IG guided)Statistical Learning (i.e. Nave Bayes)Unsupervised Learning (i.e. Nearest Neighbor)

The difficulties of computational linguistics are rooted in ambiguity .N Aux V P N

?

(Adapted from ChengXiang Zhai, CS 397cxz Fall 2003)


11/75

11


Parsing

(Adapted from ChengXiang Zhai, CS 397cxz Fall 2003)

Choose most likely parse tree

the playground

S

NP VP

BNP

N

Det

A

dog

VP PPAux V

is ona boy

chasing

NP P NP

Probabil ity of this tree=0.000015

...S

NP VP

BNP

N

dog

PPAux V

is

ona boy

chasing

NP

P NP

DetA

the playground

NP

Probabil ity of this tree=0.000011

S NP VPNP Det BNPNP BNPNP NP PPBNP NVP VVP Aux V NPVP VP PPPP P NP

V chasingAux isN dogN boyN playgroundDet theDet aP on

Grammar

Lexicon

1.0

0.30.40.3

1.0

0.01

0.003

Probabilistic CFG


12/75

12


Obstacles

Ambiguity

A man saw a boy with a telescope.

Computational Intensity

Imposes a context horizon.

Text Mining NLP ApproachLocate promising fragments using fast IR methods

(bag-of-tokens)Only apply slow NLP techniques to promising fragments

S Sh ll NLP


13/75

13


Summary: Shallow NLP

However, shallow NLP techniques are feasible and useful :Lexicon machine understandable linguistic knowledge

possible senses, definitions, synonyms, antonyms, typeof,etc.

POS Tagging limit ambiguity (word/POS), entity extractionWSD stem/synonym/hyponym matches (doc and query)

Query: Foreign cars Document: Im selling a 1976 Jaguar

Parsing logical view of information (inference?,translation?)

A man saw a boy with a telescope. Even without complete NLP, any additional knowledgeextracted from text data can only be beneficial .Ingenuity will determine the applications .

R f f NLP


14/75

14


References for NLP

C. D. Manning and H. Schutze, Foundations of Natural LanguageProcessing, MIT Press, 1999.S. Russell and P. Norvig, Artificial Intelligence: A ModernApproach, Prentice Hall, 1995.S. Chakrabarti, Mining the Web: Statistical Analysis of Hypertext

and Semi-Structured Data, Morgan Kaufmann, 2002.G. Miller, R. Beckwith, C. FellBaum, D. Gross, K. Miller, and R.Tengi. Five papers on WordNet. Princeton University, August 1993.C. Zhai, Introduction to NLP, Lecture Notes for CS 397cxz, UIUC,Fall 2003.

M. Hearst, Untangling Text Data Mining, ACL99, invited pap er.http://www.sims.berkeley.edu/~hearst/papers/acl99/acl99-tdm.htmlR. Sproat, Introduction to Computational Linguistics, LING 306,UIUC, Fall 2003.

A Road Map to Text Mining and Web Mining, University of Texa sresource page. http://www.cs.utexas.edu/users/pebronia/text-mining/Compu tational Linguistics and Text Min ing Group, IBM Research,http:// www.research.ibm.com/dssgrp/

T t D t b d I f ti R t i l
http://www.sims.berkeley.edu/~hearst/papers/acl99/acl99-tdm.htmlhttp://www.sims.berkeley.edu/~hearst/papers/acl99/acl99-tdm.htmlhttp://www.sims.berkeley.edu/~hearst/papers/acl99/acl99-tdm.htmlhttp://www.cs.utexas.edu/users/pebronia/text-mining/http://www.cs.utexas.edu/users/pebronia/text-mining/http://www.cs.utexas.edu/users/pebronia/text-mining/http://www.cs.utexas.edu/users/pebronia/text-mining/http://www.research.ibm.com/dssgrp/http://www.research.ibm.com/dssgrp/http://www.research.ibm.com/dssgrp/http://www.cs.utexas.edu/users/pebronia/text-mining/http://www.cs.utexas.edu/users/pebronia/text-mining/http://www.sims.berkeley.edu/~hearst/papers/acl99/acl99-tdm.htmlhttp://www.sims.berkeley.edu/~hearst/papers/acl99/acl99-tdm.html


15/75

15


Text Databases and Information Retrieval

Text databases (document databases)Large collections of documents from various sources:news articles, research papers, books, digital libraries,E-mail messages, and Web pages, library database, etc.Data stored is usually semi-structuredTraditional information retrieval techniques becomeinadequate for the increasingly vast amounts of text data

Information retrievalA field developed in parallel with database systemsInformation is organized into (a large number of)documentsInformation retrieval problem: locating relevantdocuments based on user input, such as keywords or

example documents

16Information Retrieval (IR)


16/75

16


Information Retrieval (IR)

Typical Information Retrieval systemsOnline library catalogsOnline document management systems

Information retrieval vs. database systemsSome DB problems are not present in IR, e.g.,update, transaction management, complex objectsSome IR problems are not addressed well in DBMS, e.g.,unstructured documents, approximate search usingkeywords and relevance

17Basic Measures for Text Retrieval


17/75

17


Basic Measures for Text Retrieval

Precision : the percentage of retrieved documents that are in factrelevant to the query (i.e., correct responses)

Recall : the percentage of documents that are relevant to the queryand were, in fact, retrieved

|}{||}|

RelevantRetrievedRelevant

precision

=

|}{||}|

RetrievedRetrievedRelevantprecision

=

Relevant Relevant & Retrieved Retrieved

All Documents

18Information Retrieval Techniques


18/75

18


Information Retrieval Techniques

Basic ConceptsA document can be described by a set of representativekeywords called index terms.Different index terms have varying relevance when usedto describe document contents.This effect is captured through the assignment of numerical weights to each index term of a document.

(e.g.: frequency, tf-idf)

DBMS AnalogyIndex Terms AttributesWeights Attribute Values

19Information Retrieval Techniques


19/75

19


Information Retrieval Techniques

Index Terms (Attribute) Selection:Stop listWord stem

Index terms weighting methods

Terms U Documents Frequency Matrices

Information Retrieval Models:Boolean ModelVector Model

Probabilistic Model

20Boolean Model


20/75

20


Boolean Model

Consider that index terms are either presentor absent in a documentAs a result, the index term weights are assumedto be all binariesA query is composed of index terms linked by threeconnectives: not , and , and or

E.g.: car and repair, plane or airplane

The Boolean model predicts that each document is eitherrelevant or non-relevant based on the match of a documentto the query

21Keyword-Based Retrieval


21/75

21


Keyword Based Retrieval

A document is represented by a string, which can beidentified by a set of keywords

Queries may use expressions of keywordsE.g., car and repair shop, tea or coffee,DBMS but not OracleQueries and retrieval should consider synonyms,

e.g., repair and maintenance

Major difficulties of the modelSynonymy: A keyword T does not appear anywhere in thedocument, even though the document is closely related toT, e.g., data miningPolysemy: The same keyword may mean different things

in different contexts, e.g., mining

22Similarity-Based Retrieval in Text Data


22/75

22


Similarity Based Retrieval in Text Data

Finds similar documents based on a set of common keywords

Answer should be based on the degree of relevance based onthe nearness of the keywords, relative frequency of thekeywords, etc.

Basic techniques

Stop listSet of words that are deemed irrelevant, even thoughthey may appear frequentlyE.g., a, the, of, for, to, with, etc.Stop lists may vary when document set varies

23Similarity-Based Retrieval in Text Data


23/75


Similarity Based Retrieval in Text Data

Word stemSeveral words are small syntactic variants of each other sincethey share a common word stemE.g., drug, drugs, drugged

A term frequency tableEach entry frequent_table(i, j) = # of occurrences of the wordti in document diUsually, the ratio instead of the absolute number of occurrencesis used

Similarity metrics: measure the closeness of a document to a query(a set of keywords)

Relative term occurrencesCosine distance:

||||),( 2121

21 vvvv

vvsim

=

24Indexing Techniques


24/75


g q

Inverted indexMaintains two hash- or B+-tree indexed tables: document_table: a set of document records term_table: a set of term records,

Answer query: Find all docs associated with one or a set of terms+ easy to implement do not handle well synonymy and polysemy, and posting listscould be too long (storage could be very large)

Signature fileAssociate a signature with each documentA signature is a representation of an ordered list of terms that

describe the documentOrder is obtained by frequency analysis, stemming and stop lists

25Vector Space Model


25/75


p

Documents and user queries are represented as m-dimensionalvectors, where m is the total number of index terms in thedocument collection.The degree of similarity of the document d with regard to the query

q is calculated as the correlation between the vectors that representthem, using measures such as the Euclidian distance or the cosineof the angle between these two vectors.

26Latent Semantic Indexing (1)


26/75


g ( )

Basic ideaSimilar documents have similar word frequenciesDifficulty: the size of the term frequency matrix is very largeUse a singular value decomposition (SVD) techniques to reduce

the size of frequency tableRetain the K most significant rows of the frequency table

Method

Create a term x document weighted frequency matrix A

SVD construction: A = U * S * V

Define K and obtain U k , , S k , and V k.

Create query vector q .

Project q into the term-document space: Dq = q * U k * S k-1

Calculate similarities: cos = Dq . D / ||Dq|| * ||D||

27Latent Semantic Indexing (2)


27/75


Weighted Frequency Matrix

Query Terms:- Insulation

- Joint

28Probabilistic Model


28/75


Basic assumption: Given a user query, there is a set of documents which contains exactly the relevant documentsand no other (ideal answer set)Querying process as a process of specifying the properties of an ideal answer set. Since these properties are not known atquery time, an initial guess is madeThis initial guess allows the generation of a preliminaryprobabilistic description of the ideal answer set which is usedto retrieve the first set of documentsAn interaction with the user is then initiated with the purposeof improving the probabilistic description of the answer set

29Types of Text Data Mining


29/75


Keyword-based association analysisAutomatic document classificationSimilarity detection

Cluster documents by a common authorCluster documents containing informationfrom a common source

Link analysis: unusual correlation between entities

Sequence analysis: predicting a recurring eventAnomaly detection: find information that violates usualpatterns

Hypertext analysisPatterns in anchors/linksAnchor text correlations with linked objects

30Keyword-Based Association Analysis


30/75


MotivationCollect sets of keywords or terms that occur frequentlytogether and then find the association or correlationrelationships among them

Association Analysis ProcessPreprocess the text data by parsing, stemming,removing stop words, etc.

Evoke association mining algorithms Consider each document as a transaction View a set of keywords in the document as

a set of items in the transaction

Term level association mining No need for human effort in tagging documents The number of meaningless results and

the execution time is greatly reduced

31Text Classification (1)


31/75


MotivationAutomatic classification for the large number of on-linetext documents (Web pages, e-mails, corporate intranets,etc.)

Classification ProcessData preprocessingDefinition of training set and test setsCreation of the classification model using the selectedclassification algorithmClassification model validationClassification of new/unknown text documents

Text document classification differsfrom the classification of relational data

Document databases are not structured according toattribute-value pairs

32Text Classification (2)


32/75


Classification Algorithms:Support Vector MachinesK-Nearest NeighborsNave BayesNeural NetworksDecision TreesAssociation rule-basedBoosting

33Document Clustering


33/75


MotivationAutomatically group related documents based on their contentsNo predetermined training sets or taxonomiesGenerate a taxonomy at runtime

Clustering ProcessData preprocessing:remove stop words, stem, feature extraction, lexical analysis,etc.Hierarchical clustering:compute similarities applying clustering algorithms.Model-Based clustering (Neural Network Approach):clusters are represented by exemplars. (e.g.: SOM)

34Text Categorization


34/75


Pre-given categories and labeled document examples(Categories may form hierarchy)Classify new documentsA standard classification (supervised learning ) problem

CategorizationSystem

Sports

Business

Education

ScienceSportsBusiness

Education

35Applications


35/75


News article classificationAutomatic email filteringWebpage classificationWord sense disambiguation

36Categorization Methods


36/75


Manual : Typically rule-basedDoes not scale up (labor-intensive, rule inconsistency)May be appropriate for special data on a particulardomain

Automatic : Typically exploiting machine learning techniquesVector space model based

Prototype-based (Rocchio) K-nearest neighbor (KNN) Decision-tree (learn rules) Neural Networks (learn non-linear classifier) Support Vector Machines (SVM)

Probabilistic or generative model based Nave Bayes classifier

37Vector Space Model


37/75


Represent a doc by a term vectorTerm: basic concept, e.g., word or phraseEach term defines one dimensionN terms define a N-dimensional spaceElement of vector corresponds to term weightE.g., d = (x 1 ,,x N), x i is importance of term I

New document is assigned to the most likely category basedon vector similarity

38Illustration of the Vector Space Model


38/75


Java

Microsoft

StarbucksC2 Category 2

C1 Category 1

C3

Category 3

new doc

39What VS Model Does Not Specify?


39/75


How to select terms to capture basic concepts Word stopping

e.g. a, the, always, along

Word stemming e.g. computer, computing, computerize => compute

Latent semantic indexingHow to assign weights

Not all words are equally important: Some are moreindicative than others

e.g. algebra vs. science

How to measure the similarity

40How to Assign Weights


40/75


Two-fold heuristics based on frequency

TF (Term frequency)More frequent within a document more relevant tosemanticsE.g., query vs. commercial

IDF (Inverse document frequency)Less frequent among documents more discriminativeE.g. algebra vs. science

41Term Frequency Weighting


41/75


WeightingThe more frequent, the more relevant to topicE.g. query vs. commercial Raw TF= f(t,d): how many times term t appears in doc d

NormalizationWhen document length varies,

then relative frequency is preferredE.g., Maximum frequency normalization

42Inverse Document Frequency Weighting


42/75


Idea: The less frequent terms among documents are themore discriminative

Formula:

Givenn, total number of docsk, the number of docs with term t appearing(the DF document frequency)

43TF-IDF Weighting


43/75


TF-IDF weighting: weight(t, d) = TF(t, d) * IDF(t)Frequent within doc high tf high weightSelective among docs high idf high weight

Recall VS modelEach selected term represents one dimensionEach doc is represented by a feature vector

Its t-term coordinate of document d is the TF-IDF weightThis is more reasonable

Just for illustration Many complex and more effective weighting variants existin practice

44How to Measure Similarity?


44/75


Given two document

Similarity definitiondot product

normalized dot product (or cosine)

45Illustrative Example


45/75


text mining travel map search engine govern president congressIDF(faked) 2.4 4.5 2.8 3.3 2.1 5.4 2.2 3.2 4.3

doc1 2(4.8) 1(4.5) 1(2.1) 1(5.4)doc2 1(2.4 ) 2 (5.6) 1(3.3)doc3 1 (2.2) 1(3.2) 1(4.3)

newdoc 1(2.4) 1(4.5)

doc3

textminingsearch

enginetext

traveltext

maptravel

governmentpresidentcongress

doc1

doc2

To whom is newdocmore similar?

Sim(newdoc,doc1)=4.8*2.4+4.5*4.5

Sim(newdoc,doc2)=2.4*2.4

Sim(newdoc,doc3)=0

46Vector Space Model-Based Classifiers


46/75


What do we have so far?A feature space with similarity measureThis is a classic supervised learning problemSearch for an approximation to classification hyper plane

VS model based classifiersK-NN

Decision tree basedNeural networksSupport vector machine

47Probabilistic Model


47/75


Category C is modeled as a probability distributionof predefined random eventsRandom events model the processof generating documents

Therefore, how likely a document d belongsto category C is measured through the probabilityfor category C to generate d.

48Quick Revisit of Bayes Rule


48/75


Category Hypothesis space: H = {C 1 , , C n}One document: D

As we want to pick the most likely category C*, we can dropp(D)

( | ) ( )( | )( )

i ii

P D C P CP C DP D

=

Posterior probability of C i

Document model for category C

* argmax ( | ) argmax ( | ) ( )C CC P C D P D C P C= =

49Probabilistic Model: Multi-Bernoulli


49/75


Event: word presence or absence

D = (x 1 , , x |V| )x i =1 for presence of word w ix i =0 for absence

Parameters

{p(wi=1|C), p(wi=0|C)}p(wi=1|C)+ p(wi=0|C)=1

| | | | | |

1 | |1 1, 1 1, 0

( ( ,..., ) | ) ( | ) ( 1| ) ( 0| )i i

V V V

V i i i ii i x i x

p D x x C p w x C p w C p w C= = = = =

= = = = = =

50Probabilistic Model: Multinomial


50/75


Event: word selection/sampling

D = (n 1 , , n |V| )n i: frequency of word w in=n 1 ,++ n |V|

Parameters

{p(wi|C)}p(w 1 |C)+ p(w |v| |C) = 1

| |

1 | |1 | | 1( ( ,..., ) | ) ( | ) ( | )...

i

Vn

v iV i

np D n n C p n C p w Cn n =

= =

51Parameter Estimation


51/75


Category prior

Multi-Bernoulli Doc model

Multinomial doc model

Training examples:

C1 C2

Ck

E(C 1)

E(C k)

E(C 2)

Vocabulary: V = {w 1 , , w |V| }

1

| ( ) |( )

| ( ) |

ii k

j j

E Cp C

E C=

=

( )( , ) 0.5

1( 1| ) ( , )

| ( )| 1 0i

j jd E C

j i ji

w dif w occursind

pw C w dEC otherwise

+

= = =+

( )| |

1 ( )

( , ) 1( | ) ( , )

( , ) | |

i

i

jd E C

j i j jV

m

m d E C

cw dpw C cw d countsof w ind

cw d V

=

+= =

+

52Classification of New Document


52/75


1 | |

| |

1| |

1

( ,..., ) {0,1}* argmax ( | ) ( )

argmax ( | ) ( )

argmax log ( ) log ( | )

V

C

V

C i i

iV

C i ii

d x x xC P D C P C

pw x C P C

pC p w x C

=

=

= =

= =

= + =

1 | | 1 | |

| |

1

| |

1

| |

1

( ,..., ) | | ...* argmax ( | ) ( )

argmax ( | ) ( | ) ( )

argmax log ( | ) log ( ) log ( | )

argmax log ( ) log ( | )

i

V V

C

Vn

C ii

V

C i ii

V

C i ii

d n n d n n nC P D C P C

pn C p w C P C

pn C pC n p w C

pC n p w C

=

=

=

= = = + +=

=

= + +

+

Multi-Bernoulli Multinomial

53Categorization Methods


53/75


Vector space modelK-NNDecision treeNeural networkSupport vector machine

Probabilistic model

Nave Bayes classifier

Many, many others and variants exist

e.g. Bim, Nb, Ind, Swap-1, LLSF, Widrow-Hoff, Rocchio,Gis-W,

54Evaluation (1)


54/75


Effectiveness measureClassic: Precision & Recall

Precision

Recall

55Evaluation (2)


55/75


BenchmarksClassic: Reuters collection A set of newswire stories classified under categories related

to economics.

EffectivenessDifficulties of strict comparison

different parameter setting different split (or selection) between training and testing various optimizations

However widely recognizable Best: Boosting-based committee classifier & SVM

Worst: Nave Bayes classifierNeed to consider other factors, especially efficiency

56Summary: Text Categorization


56/75


Wide application domainComparable effectiveness to professionalsManual Text Classification is not 100% and unlikely toimprove substantially

Automatic Text Classification is growing at a steady paceProspects and extensions

Very noisy text, such as text from O.C.R.

Speech transcripts

57Research Problems in Text Mining


57/75


Google: what is the next step?How to find the pages that match approximately thesophisticated documents, with incorporation of user-profilesor preferences?

Look back of Google: inverted indiciesConstruction of indicies for the sophisticated documents,with incorporation of user-profiles or preferencesSimilarity search of such pages using such indicies

58References for Text Mining


58/75


Fabrizio Sebastiani, Machine Learning in Automated TextCategorization, ACM Computing Surveys, Vol. 34, No.1,March 2002Soumen Chakrabarti, Data mining for hypertext: A tutorial

survey, ACM SIGKDD Explorations, 2000.Cleverdon, Optimizing convenient online accesss tobibliographic databases, Information Survey, Use4, 1, 37-47, 1984Yiming Yang, An evaluation of statistical approaches to textcategorization, Journal of Information Retrieval, 1:67-88,1999.Yiming Yang and Xin Liu A re-examination of textcategorization methods. Proceedings of ACM SIGIRConference on Research and Development in InformationRetrieval (SIGIR'99, pp 42--49), 1999.

59World Wide Web: a brief history


59/75


Who invented the wheel is unknownWho invented the World-Wide Web ?

(Sir) Tim Berners-Leein 1989, while working at CERN,invented the World Wide Web,including URL scheme, HTML, and in1990 wrote the first server and thefirst browserMosaic browser developed by MarcAndreessen and Eric Bina at NCSA

(National Center for SupercomputingApplications) in 1993; helped rapidweb spreadMosaic was basis for Netscape

60What is Web Mining?


60/75


Discovering interesting and useful informationfrom Web content and usage

ExamplesWeb search, e.g. Google, Yahoo, MSN, Ask, Specialized search: e.g. Froogle (comparison shopping),

job ads (Flipdog)eCommerceRecommendations (Netflix, Amazon, etc.)Improving conversion rate: next best product to offer

Advertising, e.g. Google AdsenseFraud detection: click fraud detection, Improving Web site design and performance

61How does it differ from classicalData Mining?


61/75


The web is not a relationTextual information and linkage structure

Usage data is huge and growing rapidlyGoogles usage logs are bigger than their web crawlData generated per day is comparable to largestconventional data warehouses

Ability to react in real-time to usage patterns

No human in the loop

Reproduced from Ullman & Rajaramanwith permission

62How big is the Web?


62/75


The number of pages is technically, infiniteBecause of dynamically generated contentLots of duplication (30-40%)

Best estimate of unique static HTML pages comes fromsearch engine claims

Google = 8 billion

Yahoo = 20 billionLots of marketing hype

Reproduced from Ullman & Rajaramanwith permission

63Netcraft Survey:76,184,000 web sites (Feb 2006)


63/75


http://news.netcraft.com/archives/web_server_survey.html

Netcraft survey

64Mining the World-Wide Web


64/75


The WWW is huge, widely distributed, global informationservice center for

Information services: news, advertisements, consumerinformation, financial management, education,

government, e-commerce, etc.Hyper-link informationAccess and usage information

WWW provides rich sources for data miningChallenges

Too huge for effective data warehousing and data miningToo complex and heterogeneous: no standards and

structure

65Web Mining: A more challenging task


65/75


Searches forWeb access patternsWeb structuresRegularity and dynamics of Web contents

ProblemsThe abundance problemLimited coverage of the Web: hidden Web sources,

majority of data in DBMSLimited query interface based on keyword-oriented searchLimited customization to individual users

66Web Mining Taxonomy


66/75


Web Mining

Web StructureMiningWeb ContentMining

Web Page

Content MiningSearch Result

Mining

Web UsageMining

General Access

Pattern Tracking

Customized

Usage Tracking



67/75


Web Mining

Web StructureMining

Web Page Content Mining

Web Page SummarizationWebLog , WebOQL :Web Structuring query languages;Can identify information within given webpagesAhoy! :Uses heuristics to distinguish personalhome pages from other web pagesShopBot : Looks for product prices within webpages

Web Content

Mining

Search ResultMining

Web UsageMining

General AccessPattern Tracking

CustomizedUsage Tracking



68/75


Web Mining

Web UsageMining



Web StructureMining

Web ContentMining

Web PageContent Mining

Search Result Mining

Search Engine Result

SummarizationClustering Search Result :Categorizes documents usingphrases in titles and snippets



69/75


Web Mining

Web Structure MiningUsing LinksPageRankCLEVERUse interconnections between web pages to giveweight to pages.

Using GeneralizationMLDB, VWVUses a multi-level database representation of theWeb. Counters (popularity) and link lists are usedfor capturing structure.

Web ContentMining


Search ResultMining

Web UsageMining





70/75


Web Mining

Web StructureMining

Web ContentMining


Search ResultMining

General Access Pattern Tracking

Web Log MiningUses KDD techniques to understandgeneral access patterns and trends.Can shed light on better structure andgrouping of resource providers.

Web UsageMining




71/75


Web Mining

Customized Usage Tracking

Adaptive Sites

Analyzes access patterns of each user at atime. Web site restructures itself automaticallyby learning from user access patterns.

Web UsageMining


Web StructureMining

Web ContentMining


Search ResultMining

72Web Usage Mining

Mi i g W b l g d t di tt f


72/75


Mining Web log records to discover user access patterns of Web pagesApplications

Target potential customers for electronic commerce

Enhance the quality and delivery of Internet informationservices to the end userImprove Web server system performanceIdentify potential prime advertisement locations

Web logs provide rich information about Web dynamicsTypical Web log entry includes the URL requested, the IPaddress from which the request originated, and a

timestamp

73Web Usage Mining

Understanding is a pre requisite to improvement


73/75


Understanding is a pre-requisite to improvement1 Google, but 70,000,000+ web sites

What Applications?

Simple and Basic Monitor performance, bandwidth usage Catch errors (404 errors- pages not found)

Improve web site design (shortcuts for frequent paths,remove links not used, etc)

Advanced and Business Critical

eCommerce: improve conversion, sales, profit Fraud detection: click stream fraud,

74Techniques for Web usage mining

Construct multidimensional view on the Weblog database


74/75


Construct multidimensional view on the Weblog databasePerform multidimensional OLAP analysis to find the top Nusers, top N accessed Web pages, most frequentlyaccessed time periods, etc.

Perform data mining on Weblog recordsFind association patterns, sequential patterns, and trendsof Web accessing

May need additional information,e.g., user browsingsequences of the Web pages in the Web server buffer

Conduct studies to

Analyze system performance, improve system design byWeb caching, Web page prefetching, and Web pageswapping

75References

Deng Cai, Shipeng Yu, Ji-Rong Wen and Wei-Ying Ma, Extracting Content Structure forb b d l h f h f b f


75/75


Web Pages based on Visual Representation, The Fifth Asia Pacific Web Conference,2003.Deng Cai, Shipeng Yu, Ji-Rong Wen and Wei-Ying Ma, VIPS: a Vision-based PageSegmentation Algorithm, Microsoft Technical Report (MSR-TR-2003-79), 2003.Shipeng Yu, Deng Cai, Ji-Rong Wen and Wei-Ying Ma, Improving Pseudo-RelevanceFeedback in Web Information Retrieval Using Web Page Segmentation, 12thInternational World Wide Web Conference (WWW2003), May 2003.

Ruihua Song, Haifeng Liu, Ji-Rong Wen and Wei-Ying Ma, Learning Block ImportanceModels for Web Pages, 13th International World Wide Web Conference (WWW2004),May 2004.Deng Cai, Shipeng Yu, Ji-Rong Wen and Wei-Ying Ma, Block-based Web Search,SIGIR 2004, July 2004 .Deng Cai, Xiaofei He, Ji-Rong Wen and Wei-Ying Ma, Block-Level Link Analysis, SIGIR

2004, July 2004 .Deng Cai, Xiaofei He, Wei-Ying Ma, Ji-Rong Wen and Hong-Jiang Zhang, OrganizingWWW Images Based on The Analysis of Page Layout and Web Link Structure, TheIEEE International Conference on Multimedia and EXPO (ICME'2004) , June 2004Deng Cai, Xiaofei He, Zhiwei Li, Wei-Ying Ma and Ji-Rong Wen, Hierarchical Clusteringof WWW Image Search Results Using Visual, Textual and Link Analysis,12th ACMInternational Conference on Multimedia, Oct. 2004 .

Machine Learning and Data Mining 19 Mining Text and Web Data 26716

Documents

Transcript of Machine Learning and Data Mining 19 Mining Text and Web Data 26716