Disambiguation and Filter Methods in Using Web Knowledge for Coreference Resolution
O. Uryupina and M. Poesio
They describe how they use Wikipedia and Yago to increase their Coreference Resolution performance. They use BART to support their efforts. Their classification consists of a anaphor and a potential antecedent, as they describe. Using their associated feature vectors, they use a 'maximum entropy classifier' to determine coreference. They used Wikipedia to improve their aliasing algorithm, which would perform string matching functions. They use Wikipedia based information as a feature, and to disambiguate mentions. They use Yago to supplement their efforts when there are too few features to make any reasonable decisions related to coreference. Yago information is also incorporated as a feature. They tested using ACE with reasonable scores.
Using these publicly available knowledge bases appears to improve performance (2-3% in this case). Something to think about....
Showing posts with label Classification. Show all posts
Showing posts with label Classification. Show all posts
Saturday, August 20, 2011
Tuesday, June 21, 2011
Canopy Clustering
"Efficient Clustering of High Dimensional Data Sets with Application to Reference Matching", McCallum,Nigam,Ungar,http://www.kamalnigam.com/papers/canopy-kdd00.pdf
This paper discusses a different type of clustering, titled canopy clustering. It is an interesting idea. There are basically two thresholds, using a 'cheap distance metric', we evaluate a list of points. Threshold 1 is > than Threshold 2. Pick one point to compare with all the other points in the list. When the distance between the two points falls within T1 put the points into a canopy. If the distance falls within T2 then remove point from list. We generate the canopies this way and work through the list until empty.
We can then apply our second level of clustering to each canopy and are pretty much guaranteed that if two points do not fall into the same canopy then they are likely not to be co-referent and therefore do not need to be evaluated.
This is efficient and elegant. Currently the only implementation that I found of canopy clustering is in Mahout. I am building my own implementation though to get a feel for how well it works.
This paper discusses a different type of clustering, titled canopy clustering. It is an interesting idea. There are basically two thresholds, using a 'cheap distance metric', we evaluate a list of points. Threshold 1 is > than Threshold 2. Pick one point to compare with all the other points in the list. When the distance between the two points falls within T1 put the points into a canopy. If the distance falls within T2 then remove point from list. We generate the canopies this way and work through the list until empty.
We can then apply our second level of clustering to each canopy and are pretty much guaranteed that if two points do not fall into the same canopy then they are likely not to be co-referent and therefore do not need to be evaluated.
This is efficient and elegant. Currently the only implementation that I found of canopy clustering is in Mahout. I am building my own implementation though to get a feel for how well it works.
SimHash: Hash-based Similarity Detection
"SimHash: Hash-based Similarity Detection", Sadowski, Levin, 2007.
This paper outlines a hash algorithm that can be used for similarity detection. Most hash algorithm are designed to offer low collision and hash values for similar strings can vary quite a bit. This hash basically sets out to achieve the opposite, higher collision and hash keys for similar strings are similar if not the same.
If you cannot use term frequency and need a numeric representation of a string for statistical processing, what process can be used? Using an integer-based hash is one way to achieve this, though in my opinion it is not the most sophisticated of approaches.
An implementation of this algorithm showed that it is a reasonable approach for hashing strings with the intent to determine similarity. I found minor issues which I will address by altering the algorithm.
Results showed that this is a reasonable approach....
This paper outlines a hash algorithm that can be used for similarity detection. Most hash algorithm are designed to offer low collision and hash values for similar strings can vary quite a bit. This hash basically sets out to achieve the opposite, higher collision and hash keys for similar strings are similar if not the same.
If you cannot use term frequency and need a numeric representation of a string for statistical processing, what process can be used? Using an integer-based hash is one way to achieve this, though in my opinion it is not the most sophisticated of approaches.
An implementation of this algorithm showed that it is a reasonable approach for hashing strings with the intent to determine similarity. I found minor issues which I will address by altering the algorithm.
Results showed that this is a reasonable approach....
Saturday, April 17, 2010
Paper Summary - An Introduction to Bayesian and Dempster-Shafer Data Fusion
D. Koks and S. Challa, DSTO Systems Sciences Laboratory, November 2005
http://www.dsto.defence.gov.au/publications/2563/DSTO-TR-1436.pdf
This paper is about data fusion and using techniques such as Bayesian and DS. It starts out with an introduction about data fusion and how it is defined in multiple domains. It then highlights work by others that implemented different methods to perform data fusion. It then gives nice detailed examples of using Bayesian and Dempster-Shafer to perform data fusion. It ends with a comparison summary of these two techniques.
The paper is very good. It shows all of the equations step by step and gives clear examples. It very clearly shows shortcoming of both methods.
http://www.dsto.defence.gov.au/publications/2563/DSTO-TR-1436.pdf
This paper is about data fusion and using techniques such as Bayesian and DS. It starts out with an introduction about data fusion and how it is defined in multiple domains. It then highlights work by others that implemented different methods to perform data fusion. It then gives nice detailed examples of using Bayesian and Dempster-Shafer to perform data fusion. It ends with a comparison summary of these two techniques.
The paper is very good. It shows all of the equations step by step and gives clear examples. It very clearly shows shortcoming of both methods.
Tuesday, March 30, 2010
Paper Summary - Link-based text classification
Q. Lu and L. Getoor, "Link-based text classification",IJCAI workshop on text-mining and link-analysis, 2003
This paper examines machine learning when objects are linked and using the links as additional information for the classifier. They present a framework which models link distributions using a logistic regression model for both content and links. They found that using links actually improved the accuracy of the classifier.
They use an iterative classification algorithm since the attributes can be correlated. There is a joint distribution between links and content attributes.
The main points in the paper are:
* The statistical framework models link distributions
* They show through results how this improved accuracy of the classifier
* They show an evaluation of the iterative categorization algorithm
This paper examines machine learning when objects are linked and using the links as additional information for the classifier. They present a framework which models link distributions using a logistic regression model for both content and links. They found that using links actually improved the accuracy of the classifier.
They use an iterative classification algorithm since the attributes can be correlated. There is a joint distribution between links and content attributes.
The main points in the paper are:
* The statistical framework models link distributions
* They show through results how this improved accuracy of the classifier
* They show an evaluation of the iterative categorization algorithm
Paper Summary - Unsupervised named-entity extraction from the web: An experimental study
"Unsupervised named-entity extraction from the web: An experimental study", O. Etzioni and M. Cafarella and D. Downey and S. Kok and A. Popescu and T. Shaked and S. Soderland and D. Weld and A. Yates, Artificial Intelligence,91-134,2005
This paper describes a system that uses an unsupervised approach at entity extraction. It describe the architecture of the system and defines general principles for extraction. They present 3 ways improve recall and extraction without compromising precision:
Pattern Learning - domain-specific
Subclass Extraction - identifies sub-classes
List Extraction - locates lists of class instances
The paper starts with a motivation for using an unsupervised approach. It provides background in information extraction and some of the complexities involved. KNOWITALL uses extraction patterns and pointwise mutal information statistics calculated from the web using hit counts measuring degree of correlation between pairs of words. The PMIs are used as features for a classifier. It consumes information from search engines.
It uses a bootstrapping method which is a set of predicates. Labels are given for each class and symbolic names are given for each class. The Bootstrapper "uses the labels to instantiate extraction rules". The Extractor formulates queries to send to search engine and the Assessor evaluates the extractions.
The paper is quite extensive and describes the system well. Overall this paper is a good read and provides a number of tidbits for new ideas.
This paper describes a system that uses an unsupervised approach at entity extraction. It describe the architecture of the system and defines general principles for extraction. They present 3 ways improve recall and extraction without compromising precision:
Pattern Learning - domain-specific
Subclass Extraction - identifies sub-classes
List Extraction - locates lists of class instances
The paper starts with a motivation for using an unsupervised approach. It provides background in information extraction and some of the complexities involved. KNOWITALL uses extraction patterns and pointwise mutal information statistics calculated from the web using hit counts measuring degree of correlation between pairs of words. The PMIs are used as features for a classifier. It consumes information from search engines.
It uses a bootstrapping method which is a set of predicates. Labels are given for each class and symbolic names are given for each class. The Bootstrapper "uses the labels to instantiate extraction rules". The Extractor formulates queries to send to search engine and the Assessor evaluates the extractions.
The paper is quite extensive and describes the system well. Overall this paper is a good read and provides a number of tidbits for new ideas.
Subscribe to:
Posts (Atom)