Upload
olivia-gomez
View
35
Download
1
Embed Size (px)
DESCRIPTION
Web Information Extraction Learning based on Probabilistic Graphical Models. Wai Lam Joint work with Tak-Lam Wong The Chinese University of Hong Kong. Introduction. Building advanced Web mining applications requires precise text information extraction a large number of different Web sites. - PowerPoint PPT Presentation
Citation preview
Web Information Extraction Learning based on Probabilistic Graphical Models
Wai Lam
Joint work with Tak-Lam Wong
The Chinese University of Hong Kong
Sept 5, 2008 The Chinese University of Hong Kong 2
Introduction
Building advanced Web mining applications requires precise text information extraction a large number of different Web sites.
Substantial human effort is needed for the information extraction task. diverse layout format content variation
Sept 5, 2008 The Chinese University of Hong Kong 3
Wrapper Adaptation Problem (1)
Sept 5, 2008 The Chinese University of Hong Kong 4
Wrapper Adaptation Problem (2)
Learnedwrapper
Wrapperlearning
Sept 5, 2008 The Chinese University of Hong Kong 5
Product Attribute Extraction and Resolution Problem (1)
The Web contains a huge number of online stores selling millions of different kinds of products.
Sept 5, 2008 The Chinese University of Hong Kong 6
Product Attribute Extraction and Resolution Problem (2)
Traditional search engines typically treat every term in a Web document in a uniform fashion. Consider the digital camera domain. Suppose a user
supplies a query: “auto white balance” trying to find cameras related to the product attribute “white balance”.
Possible results: “auto ISO” which is about “light sensitivity” different from the product attribute “white balance”
Sept 5, 2008 The Chinese University of Hong Kong 7
Product Attribute Extraction and Resolution Problem (3)
Another related desirable task is to resolve the extracted data according to their semantics.
This can improve indexing of product Web pages and support intelligent tasks such as product search or product matching.
Sept 5, 2008 The Chinese University of Hong Kong 8
Our Approach We have investigated learning frameworks for solving each
of the Web information extraction tasks just presented. Probabilistic graphical models provide a principled
paradigm harnessing the uncertainty during the learning process.
A graphical model capturing information extraction knowledge for solving wrapper adaptation (ACM TOIT 2007).
A graphical model for unsupervised learning to extract and resolve product attributes (SIGIR 2008).
Sept 5, 2008 The Chinese University of Hong Kong 9
Motivating Example
(Source: http://www.superwarehouse.com)
(Source: http://www.crayeon3.com)
Sept 5, 2008 The Chinese University of Hong Kong 10
Product Attribute Extraction
To extract product attributes: In the beginning, only the
attribute “resolution” is known. Effective sensor resolution
Layout format White balance, shutter speed
Mutual cooperation Light sensitivity
Sept 5, 2008 The Chinese University of Hong Kong 11
Product Attribute Resolution
Samples of extracted text fragments from a page: cloudy, daylight, etc… What do they refer to?
A text fragment extracted from another page: white balance auto, daylight,
cloudy, tungsten, … … Product attribute resolution:
To cluster text fragments of attributes into the same group Better indexing for product search Easier understanding and interpretation
Sept 5, 2008 The Chinese University of Hong Kong 12
Existing Works (Supervised Learning)
Supervised wrapper learning (Chang et al., IEEE TKDE 2006) They need training examples. The wrapper learned from a Web site cannot be applied
to other sites. Template-independent extraction (Zhu et al., SIGKDD
2007) They cannot handle previously unseen attributes.
Sept 5, 2008 The Chinese University of Hong Kong 13
Existing Works (Unsupervised Learning)
Handle Web pages generated from the same template (Crescenzi et al., VLDB 2001). Data may not be synchronized
• “Aug 1993 $16.38” extracted from a page
• “Paperback Feb 1985 $6.95” extracted from another page
Synchronized data extraction (Chuang et al., VLDB 2007) Requires a field model (HMM models) for each field
and it requires manually prepared training examples. Can only apply to Web pages that contain multiple records.
Sept 5, 2008 The Chinese University of Hong Kong 14
Our Framework
1. Unsupervised learning framework for jointly extracting and resolving product attributes from different Web sites (SIGIR 2008).
2. Our framework consists of a graphical model which considers page-independent content information and page-dependent layout information.
3. Can extract unlimited number of product attributes (Dirichlet process prior)
4. The resolved product attributes can be used for other intelligent tasks such as product search (AAAI 2008).
Sept 5, 2008 The Chinese University of Hong Kong 15
Problem Definition (1)
A product domain, E.g., Digital camera domain
A set of reference attributes, E.g., “resolution”, “white balance”, etc. A special element, , representing “not-an-attribute”
A collection of Web pages from any Web sites, , each of which contains a single product
Let be any text fragment from a Web page
Sept 5, 2008 The Chinese University of Hong Kong 16
Problem Definition (2)
<TR> <TD> <P> <SPAN> White balance </SPAN> </P> </TD> <TD> <P> <SPAN> Auto, daylight, cloudy, tungstem, fluorescent, fluorescent H, custom </SPAN> </P> </TD></TR><TR>
<TR> <TD> <P> <SPAN> White balance </SPAN> </P> </TD> <TD> <P> <SPAN> Auto, daylight, cloudy, tungstem, fluorescent, fluorescent H, custom </SPAN> </P> </TD></TR><TR>
Line separator
Line separator
Sept 5, 2008 The Chinese University of Hong Kong 17
Problem Definition (3)
Attribute information
Target informationLayout information
Content information
White balance Auto, daylight, … …
boldface, in-table
1 (related to attribute)
white balance
Sept 5, 2008 The Chinese University of Hong Kong 18
Problem Definition (4)
Attribute information
Target informationLayout information
Content information
View larger image
boldface, underline
0 (irrelevant)
not-an-attribute
Sept 5, 2008 The Chinese University of Hong Kong 19
Attribute extraction:
Attribute resolution:
Joint attribute extraction and resolution:
Problem Definition (5)
Attribute information
Target informationLayout information
Content information
Sept 5, 2008 The Chinese University of Hong Kong 20
Graphical Models (1)
A graphical model is a family of probability distributions defined in terms of a directed or undirected graph. Nodes: Random variables Joint distribution: The products over functions defined
on the connected nodes It provides general algorithms to compute marginal
and conditional probability of interest. It provides control over the computational
complexity associated with these operations.
Sept 5, 2008 The Chinese University of Hong Kong 21
Graphical Models (2)
One kind of graphical models is directed graph. Let be a directed acyclic graph
are the nodes are the edges
Denote as the parents of . Denote as the collection of random
variables indexed by the nodes. The joint probability distribution is expressed as:
Sept 5, 2008 The Chinese University of Hong Kong 22
Graphical Models (3)
E.g.:
This model asserts that the variables ZN are conditionally independent and identically distributed given θ.
Z1 Z2 Z3 ZN
θ
Sept 5, 2008 The Chinese University of Hong Kong 23
Graphical Models (4)
A plate is used to show the repetition of the variables. Hence, it shows the factorial and nested structures.
Zn
θ
N
Sept 5, 2008 The Chinese University of Hong Kong 24
Graphical Models (5)
A generative approach to clustering: pick one of clusters from a distribution generate a data point from a cluster-specific probability
distribution. This yields a finite mixture model:
where and are the parameters, and where each cluster has the same parameterized family.
Data are assumed to be generated conditionally IID from this mixture.
Finite Mixture Model
Sept 5, 2008 The Chinese University of Hong Kong 25
Graphical Models (6)
Mixture models make the assumption that each data point arises from a single mixture component. the k-th cluster is by definition the set of data points ar
ising from the k-th mixture component.
Finite Mixture Model
Sept 5, 2008 The Chinese University of Hong Kong 26
Graphical Models (7)
Another way to express this: define an underlying measure
where is an atom at . And define the process of obtaining a sample from a finite mix
ture model as follows. For :
Note that each is equal to one of the underlying . indeed, the subset of that maps to is exactly the k-th clu
ster.
Finite Mixture Model
Sept 5, 2008 The Chinese University of Hong Kong 27
Graphical Models (8)
θi
N
xi
G
Finite Mixture Model
Sept 5, 2008 The Chinese University of Hong Kong 28
Graphical Models (9)
Define a countably infinite mixture model by taking K to infinity and hoping that means something, where
Dirichlet Process Mixture
πk
ψkG0
Zi
Nxi
α
Sept 5, 2008 The Chinese University of Hong Kong 29
Our Model (1)
Our graphical model can be regarded as an extension of Dirichlet mixture model.
Each mixture component refers to a reference attribute; consists of two distributions characterizing the content
information and target information. Dirichlet process prior is employed.
It can handle unlimited number of reference attributes.
Sept 5, 2008 The Chinese University of Hong Kong 30
Attribute extraction:
Attribute resolution:
Joint attribute extraction and resolution:
Our Model (2)
Attribute information
Target informationLayout information
Content information
Sept 5, 2008 The Chinese University of Hong Kong 31
Our Model (3)Dirichlet Process Prior(Infinite Mixture Model) N Text Fragment S Different Web Site
Sept 5, 2008 The Chinese University of Hong Kong 32
Our Model (4)
N Text Fragment
Target information
Layout information
Content information
Dirichlet Process Prior(Infinite Mixture Model)
The proportion ofthe k-th component in the mixture
Content information parameterof the k-th component
Target information parameterof the k-th component
Sept 5, 2008 The Chinese University of Hong Kong 33
Our Model (5)
S Different Web Site
Site-dependent Layout format
Sept 5, 2008 The Chinese University of Hong Kong 34
Our Model (6)Dirichlet Process Prior(Infinite Mixture Model)
Concentration parameter for DP
Base distribution for content info.
Base distribution for target info.
Sept 5, 2008 The Chinese University of Hong Kong 35
Generation Process (1)
Sept 5, 2008 The Chinese University of Hong Kong 36
Generation Process (2)
The joint probability for generating a particular text fragment given the parameters, , , , and, :
Inference:
where , , and are the set of observable variables, unobservable variables, and model parameters respectively.
Intractable
Sept 5, 2008 The Chinese University of Hong Kong 37
Variational Method (1)
The inference problem is transformed into an optimization problem.
The resulting variational optimization problems admit principled approximate solutions.
The solution to variational problems is often given in terms of fixed point equations that capture necessary conditions for optimality.
In contrast to other approximation methods such as MCMC, variational methods are deterministic.
Sept 5, 2008 The Chinese University of Hong Kong 38
Variational Method (2)
Finding is intractable Our goal: Transform the problem into an
optimization problem:
where D denotes KL-divergence KL-divergence must be non-negative
Sept 5, 2008 The Chinese University of Hong Kong 39
Variational Method (3)
KL-divergence is zero if equals the true posterior probability . Let By maximizing w.r.t. we get:
Therefore, we have a lower bound on the desired log-marginal probability
LHS is the log-likelihood of the observable variables. .
Sept 5, 2008 The Chinese University of Hong Kong 40
Variational Method (4)
The problem becomes maximizing .
Sept 5, 2008 The Chinese University of Hong Kong 41
Variational Method (5)
Truncated stick-breaking process (Ishwaran and James, 2001) Replace infinity with a truncation level K
Sept 5, 2008 The Chinese University of Hong Kong 42
Variational Method (6)
Mixture of tokens
Binary
A set of binary featuresConjugate priors
Sept 5, 2008 The Chinese University of Hong Kong 43
Variational Method (7)
Solve by coordinate ascent algorithm One important variational parameters:
How likely does come from the k-th component? Attribute resolution!
Sept 5, 2008 The Chinese University of Hong Kong 44
Variational Method (8)
Another important variational parameter:
where
How likely should be extracted? Attribute extraction!
Sept 5, 2008 The Chinese University of Hong Kong 45
Variational Method (9)
Other variational parameters:
Sept 5, 2008 The Chinese University of Hong Kong 46
Initialization
What should be extracted? Make use of a very small amount of prior
information about a domain. Only a few terms about the product attributes
• E.g., resolution, light sensitivity
Can be easily obtained, for example, by just highlighting the attributes of one single Web page
Initialization
Sept 5, 2008 The Chinese University of Hong Kong 47
EM Algorithm for Layout Parameters
Our framework can consider the page-dependent layout format of text fragments to enhance extraction.
However, the layout information of an unseen Web page is unknown and hence we cannot predefine or estimate the values of .
E-step:Apply coordinate ascent algorithm until convergence to achieve the optimal conditions for all variational parameters.
M-step:Calculate
Sept 5, 2008 The Chinese University of Hong Kong 48
Experiments
We have conducted experiments on four different domains: Digital camera: 85 Web pages from 41 different sites MP3 player: 96 Web pages from 62 different sites Camcorder: 111 Web pages from 61 different sites Restaurant: 29 Web pages from LA-Weekly Restaurant
Guide
In each domain, we conducted 10 runs of experiments. In each run, we randomly selected a Web page and
pick a few terms inside for initialization.
Sept 5, 2008 The Chinese University of Hong Kong 49
Evaluation on Attribute Resolution
Baseline approach (Bilenko & Mooney SIGKDD 2003):
Agglomerative clustering Edit distance between text fragments
Evaluation metrics: Pairwise recall (R) Pairwise precision (P) Pairwise F1-measure (F)
Sept 5, 2008 The Chinese University of Hong Kong 50
Results of Attribute Resolution
Sept 5, 2008 The Chinese University of Hong Kong 51
Visualize the Resolved Attributes
The top five weighted terms in the ten largest resolved attributes in the digital camera domain:
Sept 5, 2008 The Chinese University of Hong Kong 52
Evaluation on Attribute Extraction
Surprisingly, in the restaurant domain, our framework achieves a performance (0.95 F1-measure) which is comparable to the supervised method (Muslea et al. 2001)
Sept 5, 2008 The Chinese University of Hong Kong 53
Conclusions
We investigate learning frameworks automating and adapting the extraction task based on probabilistic graphical models which provide a principled paradigm harnessing the uncertainty during the learning process.
We have developed a graphical model, which employs Dirichlet process prior, to model the generation of text fragments in Web pages for solving the tasks of product attribute extraction and resolution from different Web sites.
An unsupervised inference algorithm based on variational method is designed.
We formally show that content and layout information can collaborate and improve both extraction and resolution performance under our model.
Questions and Answers
Sept 5, 2008 The Chinese University of Hong Kong 55
Variational Method (1)
Finding is intractable Our goal: Transform the problem into an
optimization problem:
Since KL divergence must be non-negative
LHS is the log-likelihood of the observable variables
Sept 5, 2008 The Chinese University of Hong Kong 56
Variational Method (2)
KL divergence:
The problem becomes maximizing
Sept 5, 2008 The Chinese University of Hong Kong 57
Variational Method (3)
Truncated stick-breaking process (Ishwaran and James, 2001) Replace infinity with a truncation level K
Sept 5, 2008 The Chinese University of Hong Kong 58
Variational Inference (4)
Mixture of tokens
Binary
A set of binary featuresConjugate priors
Sept 5, 2008 The Chinese University of Hong Kong 59
Variational Method (5)
After applying the truncated stick-breaking process:
Sept 5, 2008 The Chinese University of Hong Kong 60
Variational Method (6)
Solve by coordinate ascent. Differentiate the formula and set to zero:
Sept 5, 2008 The Chinese University of Hong Kong 61
Variational Method (7)
One important variational parameters:
How likely does come from the k-th component? Attribute resolution!
Sept 5, 2008 The Chinese University of Hong Kong 62
Variational Method (8)
Another important variational parameter:
where
How likely should be extracted? Attribute extraction!
Sept 5, 2008 The Chinese University of Hong Kong 63
Unsupervised Approach
We make use of the prior knowledge, which is in the form of a list of a few terms, denoted as , related to product attributes.
Let be the i-th term in the list. The terms are not required to be categorized into different attributes. For each ,we select the i-th component in our model and set a high
er value of if is equal to the , and zero otherwise. In particular, we set to 10 for such . Next, for these components, we set and . This essentially means that 6 out of 10 text fragments in this compon
ent will be a text fragment related to attribute values. and for other components.