Active Inference for Collective Classification.
Labeling nodes in a network is an important problem that has seen a growing interest. A number of methods that exploit both local and relational information have been developed for this task. Acquiring the labels for a few nodes at inference time can greatly improve the accuracy, however the question of figuring out which node labels to acquire is challenging. Previous approaches have been based on simple structural properties. Here, we present a novel technique, which we refer to as reflect and correct, that can learn and predict when the underlying classification system is likely to make mistakes and it suggests acquisitions to correct those mistakes.
Learning Algorithms for Link Prediction Based on Chance Constraints.
Abstract. In this paper, we consider the link prediction problem, where we are given a partial snapshot of a network at some time and the goal is to predict the additional links formed at a later time. The accuracy of current prediction methods is quite low due to the extreme class skew and the large number of potential links. Here, we describe learning algorithms based on chance constrained programs and show that they exhibit all the properties needed for a good link predictor, namely, they allow preferential bias to positive or negative class; handle skewness in the data; and scale to large networks. Our experimental results on three real-world domains—co-authorship networks, biological networks and citation networks—show significant performance improvement over baseline algorithms. We conclude by briefly describing some promising future directions based on this work. 1
Active Learning for Networked Data.
We introduce a novel active learning algorithm for classification of network data. In this setting, training instances are connected by a set of links to form a network, the labels of linked nodes are correlated, and the goal is to exploit these dependencies and accurately label the nodes. This problem arises in many domains, including social and biological network analysis and document classification, and there has been much recent interest in methods that collectively classify the nodes in the network. While in many cases labeled examples are expensive, often network information is available. We show how an active learning algorithm can take advantage of network structure. Our algorithm effectively exploits the links between instances and the interaction between the local and collective aspects of a classifier to improve the accuracy of learning from fewer labeled examples. We experiment with two real-world benchmark collective classification domains, and show that we are able to achieve extremely accurate results even when only a small fraction of the data is labeled. 1.
Graph Identification.

Growing a tree in the forest constructing folksonomies by integrating structured metadata.
Many social Web sites allow users to annotate the content with descriptive metadata, such as tags, and more recently to organize content hierarchically. These types of structured metadata provide valuable evidence for learning how a community organizes knowledge. For instance, we can aggregate many personal hierarchies into a common taxonomy, also known as a folksonomy, that will aid users in visualizing and browsing social content, and also to help them in organizing their own content. However, learning from social metadata presents several challenges, since it is sparse, shallow, ambiguous, noisy, and inconsistent. We describe an approach to folksonomy learning based on relational clustering, which exploits structured metadata contained in personal hierarchies. Our approach clusters similar hierarchies using their structure and tag statistics, then incrementally weaves them into a deeper, bushier tree. We study folksonomy learning using social metadata extracted from the photosharing site Flickr, and demonstrate that the proposed approach addresses the challenges. Moreover, comparing to previous work, the approach produces larger, more accurate folksonomies, and in addition, scales better.
Constructing folksonomies by integrating structured metadata.
Many social Web sites allow users to annotate the content with descriptive metadata, such as tags, and more recently to organize content hierarchically. These types of structured metadata provide valuable evidence for learning how a community organizes knowledge. For instance, we can aggregate many personal hierarchies into a common taxonomy, also known as a folksonomy, that will aid users in visualizing and browsing social content, and also to help them in organizing their own content. However, learning from social metadata presents several challenges, since it is sparse, shallow, ambiguous, noisy, and inconsistent. We describe an approach to folksonomy learning based on relational clustering, which exploits structured metadata contained in personal hierarchies. Our approach clusters similar hierarchies using their structure and tag statistics, then incrementally weaves them into a deeper, bushier tree. We study folksonomy learning using social metadata extracted from the photosharing site Flickr, and demonstrate that the proposed approach addresses the challenges. Moreover, comparing to previous work, the approach produces larger, more accurate folksonomies, and in addition, scales better.
AI Theory and Practice A Discussion on Hard Challenges and Opportunities Ahead.

Indirect two%2Dsided relative ranking a robust similarity measure for gene expression data.

Integrating Structured Metadata with Relational Affinity Propagation



A Probabilistic Approach for Learning Folksonomies from Structured Data

Read%2DOnce Functions and Query Evaluation in Probabilistic Databases.
Probabilistic databases hold promise of being a viable means for large-scale uncertainty management, increasingly needed in a number of real world applications domains. However, query evaluation in probabilistic databases remains a computational challenge. Prior work on efficient exact query evaluation in probabilistic databases has largely concentrated on query-centric formulations (e.g., safe plans, hierarchical queries), in that, they only consider characteristics of the query and not the data in the database. It is easy to construct examples where a supposedly hard query run on an appropriate database gives rise to a tractable query evaluation problem. In this paper, we develop efficient query evaluation techniques that leverage characteristics of both the query and the data in the database. We focus on tuple-independent databases where the query evaluation problem is equivalent to computing marginal probabilities of Boolean formulas associated with the result tuples. Query evaluation is easy if the Boolean formulas can be factorized into a form that has every variable appearing at most once (called read-once); this suggests a naive approach that incorporates previously developed Boolean formula factorization algorithms into the query evaluation. We then develop novel, more efficient factorization algorithms that work for a large subclass of queries (specifically, conjunctive queries without self-joins), by exploiting the unique structure of the result tuple Boolean formulas. We empirically demonstrate that our proposed techniques are (1) orders of magnitude faster than generic inference algorithms when used to evaluate general read-once functions, and (2) for the special case of hierarchical queries, they rival the efficiency of prior techniques specifically designed to handle such queries. 1.
Identifying graphs from noisy and incomplete data.
There is a growing wealth of data describing networks of various types, including social networks, physical networks such as transportation or communication networks, and biological networks. At the same time, there is a growing interest in analyzing these networks, in order to uncover general laws that govern their structure and evolution, and patterns and predictive models to develop better policies and practices. However, a fundamental challenge in dealing with this newly available observational data describing networks is that the data is often of dubious quality – it is noisy and incomplete – and before any analysis method can be applied, the data must be cleaned, and missing information inferred. In this paper, we introduce the notion of graph identification, which explicitly models the inference of a “cleaned ” output network from a noisy input graph. It is this output network that is appropriate for further analysis. We present an illustrative example and use the example to explore the types of inferences involved in graph identification, as well as the challenges and issues involved in combining those inferences. We then present a simple, general approach to combining the inferences in graph identification and experimentally show the utility of our combined approach and how the performance of graph identification is sensitive to the inter-dependencies among these inferences.
Proceedings of the 1st ACM SIGKDD Workshop on Knowledge Discovery from Uncertain Data

The Dynamics of Actor Loyalty to Groups in Affiliation Networks.

The Dynamics of Actor Loyalty to Groups in Affiliation Networks.

Supervised and Unsupervised Methods in Employing Discourse Relations for Improving Opinion Polarity Classification.
This work investigates design choices in modeling a discourse scheme for improving opinion polarity classification. For this, two diverse global inference paradigms are used: a supervised collective classification framework and an unsupervised optimization framework. Both approaches perform substantially better than baseline approaches, establishing the efficacy of the methods and the underlying discourse scheme. We also present quantitative and qualitative analyses showing how the improvements are achieved. 1
Distinguishing Knowledge vs Social Capital in Social Media with Roles and Context.
Social media communities (e.g. Wikipedia, Flickr, Live Q&amp;A) give rise to distinct types of content, foremost among which are relational content (discussion, chat) and factual content (answering questions, problem-solving). Both users and researchers are increasingly interested in developing strategies that can rapidly distinguish these types of content. While many text-based and structural strategies are possible, we extend two bodies of research that show how social context, and the social roles of answerers can predict content type. We test our framework on a dataset of manually labeled contributions to Microsoft&#039;s Live Q&amp;A and find that it reliably extracts factual and relational messages from the data. 1.
Co%2Devolution of social and affiliation networks.
In our work, we address the problem of modeling social network generation which explains both link and group formation. Recent studies on social network evolution propose generative models which capture the statistical properties of real-world networks related only to node-to-node link formation. We propose a novel model which captures the coevolution of social and affiliation networks. We provide surprising insights into group formation based on observations in several real-world networks, showing that users often join groups for reasons other than their friends. Our experiments show that the model is able to capture both the newly observed and previously studied network properties. This work is the first to propose a generative model which captures the statistical properties of these complex networks. The proposed model facilitates controlled experiments which study the effect of actors ’ behavior on the network evolution, and it allows the generation of realistic synthetic datasets.
Identifying graphs from noisy and incomplete data.
There is a growing wealth of data describing networks of various types, including social networks, physical networks such as transportation or communication networks, and biological networks. At the same time, there is a growing interest in analyzing these networks, in order to uncover general laws that govern their structure and evolution, and patterns and predictive models to develop better policies and practices. However, a fundamental challenge in dealing with this newly available observational data describing networks is that the data is often of dubious quality – it is noisy and incomplete – and before any analysis method can be applied, the data must be cleaned, and missing information inferred. In this paper, we introduce the notion of graph identification, which explicitly models the inference of a “cleaned ” output network from a noisy input graph. It is this output network that is appropriate for further analysis. We present an illustrative example and use the example to explore the types of inferences involved in graph identification, as well as the challenges and issues involved in combining those inferences. We then present a simple, general approach to combining the inferences in graph identification and experimentally show the utility of our combined approach and how the performance of graph identification is sensitive to the inter-dependencies among these inferences.
FutureRank Ranking Scientific Articles by Predicting their Future PageRank.

FutureRank Ranking Scientific Articles by Predicting their Future PageRank.

On Maximum Coverage in the Streaming Model %26 Application to Multi%2Dtopic Blog%2DWatch.
We generalize the graph streaming model to hypergraphs. In this streaming model, hyperedges are arriving online and any computation has to be done on-the-fly using a small amount of space. Each hyperedge can be viewed as a set of elements (nodes), so we refer to our proposed model as the “set-streaming ” model of computation. We consider the problem of “maximum coverage”, in which k sets have to be selected that maximize the total weight of the covered elements. In the set-streaming model of computation, we show that our algorithm for maximumcoverage achieves an approximation factor of 1 4. When multiple passes are allowed, we also provide a Θ(log n) approximation algorithm for the set-cover. We next consider a multi-topic blog-watch application, an extension of blogalert like applications for handling simultaneous multipletopic requests. We show how the problems of maximumcoverage and set-cover in the set-streaming model can be utilized to give efficient online solutions to this problem. We verify the effectiveness of our methods both on synthetic and real weblog data. 1
To join or not to join the illusion of privacy in social networks with mixed public and private user profiles.
In order to address privacy concerns, many social media websites allow users to hide their personal profiles from the public. In this work, we show how an adversary can exploit an online social network with a mixture of public and private user profiles to predict the private attributes of users. We map this problem to a relational classification problem and we propose practical models that use friendship and group membership information (which is often not hidden) to infer sensitive attributes. The key novel idea is that in addition to friendship links, groups can be carriers of significant information. We show that on several well-known social media sites, we can easily and accurately recover the information of private-profile users. To the best of our knowledge, this is the first work that uses link-based and group-based classification to study privacy implications in social networks with mixed public and private user profiles.
Index Interactions in Physical Design Tuning Modeling
One of the key tasks of a database administrator is to optimize the set of materialized indices with respect to the current workload. To aid administrators in this challenging task, commercial DBMSs provide advisors that recommend a set of indices based on a sample workload. It is left for the administrator to decide which of the recommended indices to materialize and when. This decision requires some knowledge of how the indices benefit the workload, which may be difficult to understand if there are any dependencies or interactions among indices. Unfortunately, advisors do not provide this crucial information as part of the recommendation. Motivated by this shortcoming, we propose a framework and associated tools that can help an administrator understand the interactions within the recommended set of indices. We formalize the notion of index interactions and develop a novel algorithm to identify the interaction relationships that exist within a set of indices. We present experimental results with a prototype implementation over IBM DB2 that demonstrate the efficiency of our approach. We also describe two new database tuning tools that utilize information about index interactions. The first tool visualizes interactions based on a partitioning of the index-set into non-interacting subsets, and the second tool computes a schedule that materializes the indices over several maintenance windows with maximal overall benefit. In both cases, we provide strong analytical results showing that index interactions can enable enhanced functionality. 1.
Summary of the first ACM SIGKDD workshop on knowledge discovery from uncertain data %28U%2709%29.

PrDB managing and exploiting rich correlations in probabilistic databases.

Reflect and correct A misclassification prediction approach to active inference.

PrDB managing and exploiting rich correlations in probabilistic databases.

Probabilistic
Abstract. This paper presents a Bayesian method for constructing probabilistic networks from databases. In particular, we focus on constructing Bayesian belief networks. Potential applications include computer-assisted hypothesis testing, automated scientific discovery, and automated construction of probabilistic expert systems. We extend the basic method to handle missing data and hidden (latent) variables. We show how to perform probabilistic inference by averaging over the inferences of multiple belief networks. Results are presented of a preliminary evaluation of an algorithm for constructing a belief network from a database of cases. Finally, we relate the methods in this paper to previous work, and we discuss open problems.
Effective label acquisition for collective classification.

Using Friendship Ties and Family Circles for Link Prediction.

Leveraging social context for searching social media.

Trusting spam reporters A reporter%2Dbased reputation system for email filtering.

Collective Classification in Network Data.
■ Many real-world applications produce networked data such as the worldwide web (hypertext documents connected through hyperlinks), social networks (such as people connected by friendship links), communication networks (computers connected through communication links), and biological networks (such as protein interaction networks). A recent focus in machine-learning research has been to extend traditional machine-learning classification techniques to classify nodes in such networks. In this article, we provide a brief introduction to this area of research
Preface.
part of a series of studies in support of the California Governor’s Advisory Committee on Education Excellence. i PREFACE i LIST OF FIGURES iv
Learning Structured Bayesian Networks Combining Abstraction Hierarchies and Tree%2DStructured Conditional Probability Tables.

Cost%2Dsensitive learning with conditional Markov networks.

Interactive Entity Resolution in Relational Data A Visual Analytic Tool and Its Evaluation.

Structured machine learning the next ten years.
  The field of inductive logic programming (ILP) has made steady progress, since the first ILP workshop in 1991, based on a balance of developments in theory, implementations and applications. More recently there has been an increased emphasis on Probabilistic ILP and the related fields of Statistical Relational Learning (SRL) and Structured Prediction. The goal of the current paper is to consider these emerging trends and chart out the strategic directions and open problems for the broader area of structured machine learning for the next 10 years.
Exploiting shared correlations in probabilistic databases.
There has been a recent surge in work in probabilistic databases, propelled in large part by the huge increase in noisy data sources — from sensor data, experimental data, data from uncurated sources, and many others. There is a growing need for database management systems that can efficiently represent and query such data. In this work, we show how data characteristics can be leveraged to make the query evaluation process more efficient. In particular, we exploit what we refer to as shared correlations where the same uncertainties and correlations occur repeatedly in the data. Shared correlations occur mainly due to two reasons: (1) Uncertainty and correlations usually come from general statistics and rarely vary on a tuple-to-tuple basis; (2) The query evaluation procedure itself tends to re-introduce the same correlations. Prior work has shown that the query evaluation problem on probabilistic databases is equivalent to a probabilistic inference problem on an appropriately constructed probabilistic graphical model (PGM). We leverage this by introducing a new data structure, called the random variable elimination graph (rv-elim graph) that can be built from the PGM obtained from query evaluation. We develop techniques based on bisimulation that can be used to compress the rv-elim graph exploiting the presence of shared correlations in the PGM, the compressed rv-elim graph can then be used to run inference. We validate our methods by evaluating them empirically and show that even with a few shared correlations significant speed-ups are possible. 1.
VOILA Efficient Feature%2Dvalue Acquisition for Classification.

Online Collective Entity Resolution.

Relationship Identification for Social Network Discovery.
In recent years, informal, online communication has transformed the ways in which we connect and collaborate with friends and colleagues. With millions of individuals communicating online each day, we have a unique opportunity to observe the formation and evolution of roles and relationships in networked groups and organizations. Yet a number of challenges arise when attempting to infer the underlying social network from data that is often ambiguous, incomplete and context-dependent. In this paper, we consider the problem of collaborative network discovery from domains such as intelligence analysis and litigation support where the analyst is attempting to construct a validated representation of the social network. We specifically address the challenge of relationship identification where the objective is to identify relevant communications that substantiate a given social relationship type. We propose a supervised ranking approach to the problem and assess its performance on a manager-subordinate relationship identification task using the Enron email corpus. By exploiting message content, the ranker routinely cues the analyst to relevant communications relationships and message traffic that are indicative of the social relationship.
A dual%2Dview approach to interactive network visualization.
Visualizing network data, from tree structures to arbitrarily connected graphs, is a difficult problem in information visualization. A large part of the problem is that in network data, users not only have to visualize the attributes specific to each data item, but also the links specifying how those items are connected to each other. Past approaches to resolving these difficulties focus on zooming, clustering, filtering and applying various methods of laying out nodes and edges. Such approaches, however, focus only on optimizing a network visualization in a single view, limiting the amount of information that can be shown and explored in parallel. Moreover, past approaches do not allow users to cross reference different subsets or aspects of large, complex networks with each other. In this paper, we propose an approach which uses multiple coordinated views of a given network. To illustrate our approach, we implement a tool called DualNet and evaluate the tool with a case study using an email communication network. We show how using multiple coordinated views improves navigation and provides insight into large networks with multiple node and link types.
Combining Collective Classification and Link Prediction.

Representing Tuple and Attribute Uncertainty in Probabilistic Databases.
There has been a recent surge in work in probabilistic databases, propelled in large part by the huge increase in noisy data sources — sensor data, experimental data, data from uncurated sources, and many others. There is a growing need to be able to flexibly represent the uncertainties in the data, and to efficiently query the data. Building on existing probabilistic database work, we present a unifying framework which allows a flexible representation of correlated tuple and attribute level uncertainties. An important capability of our representation is the ability to represent shared correlation structures in the data. We provide motivating examples to illustrate when such shared correlation structures are likely to exist. Representing shared correlations structures allows the use of sophisticated inference techniques based on lifted probabilistic inference that, in turn, allows us to achieve significant speedups while computing probabilities for results of user-submitted queries. 1
Characterizing RNA Secondary%2DStructure Features and Their Effects on Splice%2DSite Prediction.
Abstract — RNA molecules are distinguished by their sequence composition and by their three-dimensional shape, called the secondary structure. The secondary structure of a pre-mRNA sequence may have a strong influence on gene splicing. In our previous work, we showed that a splice-site model employing sequence features built using our feature generation algorithm was very effective in predicting splice sites. The generated sequence features also contained biologically relevant features. In this paper, we extend the feature generation algorithm to construct secondary-structure features. These features capture the nucleotide pairing tendency in the splice-site neighborhood. We extend the splice-site model to include both pre-mRNA sequence and structure characteristics. The new model significantly outperforms the sequence-based features model. The identified secondary-structure features capture biologically relevant signals such as splicing silencers. We also found these signals to prefer specific regions around the splice-site neighborhood and we detail their preference. I.
GeoDDupe A Novel Interface for Interactive Entity Resolution in Geospatial Data.
Due to the growing interest in geospatial data mining and analysis, data cleaning and integration in geospatial data is becoming an important issue. Geospatial entity resolution is the process of reconciling multiple location references to the same real world location within a single data source (deduplication) or across multiple data sources (integration). In this paper, we introduce an interactive tool called GeoDDupe which effectively combines automatic data mining algorithms for geospatial entity resolution with a novel network visualization supporting users ’ resolution analysis and decisions. We illustrate the GeoDDupe interface with an example geospatial dataset and show how users can efficiently and accurately resolve location entities. Finally, the case study with two real-world geospatial datasets demonstrates the potential of GeoDDupe. 1.
Visual Mining of Multi%2DModal Social Networks at Different Abstraction Levels.
Social networks continue to become more and more feature rich. Using local and global structural properties and descriptive attributes are necessary for more sophisticated social network analysis and support for visual mining tasks. While a number of visualization tools for social network applications have been developed, most of them are limited to uni-modal graph representations. Some of the tools support a wide range of visualization options, including interactive views. Others have better support for calculating structural graph properties such as the density of the graph or deploying traditional statistical social network analysis. We present Invenio, a new tool for visual mining of socials. Invenio integrates a wide range of interactive visualization options from Prefuse, with graph mining algorithm support from JUNG. While the integration expands the breadth of functionality within the core engine of the tool, our goal is to interactively explore multi-modal, multi-relational social networks. Invenio also supports construction of views using both database operations and basic graph mining operations. 1
Graph Identification.

07161 Abstracts Collection %2D%2D Probabilistic

Preserving the Privacy of Sensitive Relationships in Graph Data.
Abstract. In this paper, we focus on the problem of preserving the privacy of sensitive relationships in graph data. We refer to the problem of inferring sensitive relationships from anonymized graph data as link reidentification. We propose five different privacy preservation strategies, which vary in terms of the amount of data removed (and hence their utility) and the amount of privacy preserved. We assume the adversary has an accurate predictive model for links, and we show experimentally the success of different link re-identification strategies under varying structural characteristics of the data.
07161 Abstracts Collection %2D%2D Probabilistic

Leveraging data and structure in ontology integration.

Probabilistic interval XML.
Abstract. Interest in XML databases has been growing over the last few years. In this paper, we study the problem of incorporating probabilistic information into XML databases. We propose the Probabilistic Interval XML (PIXml for short) data model in this paper. Using this data model, users can express probabilistic information within XML markups. In addition, we provide two alternative formal model-theoretic semantics for PIXml data. The first semantics is a “global ” semantics which is relatively intuitive, but is not directly amenable to computation. The second semantics is a “local ” semantics which is more amenable to computation. We prove several results linking the two semantics together. To our knowledge, this is the first formal model theoretic semantics for probabilistic interval XML. We then provide an operational semantics that may be used to compute answers to queries and that is correct for a large class of probabilistic instances. 1
Features generated for computational splice%2Dsite prediction correspond to functional elements.

Increasing the Predictive Power of Affiliation Networks.

Query%2Dtime Entity Resolution.
Entity resolution is the problem of reconciling database references corresponding to the same real-world entities. Given the abundance of publicly available databases that have unresolved entities, we motivate the problem of query-time entity resolution: quick and accurate resolution for answering queries over such ‘unclean ’ databases at query-time. Since collective entity resolution approaches — where related references are resolved jointly — have been shown to be more accurate than independent attribute-based resolution for off-line entity resolution, we focus on developing new algorithms for collective resolution for answering entity resolution queries at query-time. For this purpose, we first formally show that, for collective resolution, precision and recall for individual entities follow a geometric progression as neighbors at increasing distances are considered. Unfolding this progression leads naturally to a two stage ‘expand and resolve ’ query processing strategy. In this strategy, we first extract the related records for a query using two novel expansion operators, and then resolve the extracted records collectively. We then show how the same strategy can be adapted for query-time entity resolution by identifying and resolving only those database references that are the most helpful for processing the query. We validate our approach on two large real-world publication databases where we show the usefulness of collective resolution and at the same time demonstrate the need for adaptive strategies for query processing. We then show how the same queries can be answered in real-time using our adaptive approach while preserving the gains of collective resolution. In addition to experiments on real datasets, we use synthetically generated data to empirically demonstrate the validity of the performance trends predicted by our analysis of collective entity resolution over a wide range of structural characteristics in the data. 1.
SplicePort %2D An interactive splice%2Dsite analysis tool.

Visual analysis of dynamic group membership in temporal social networks.
C-Group is a tool for analyzing dynamic group membership in temporal social networks over time. Unlike most network visualization tools, which show the group structure within an entire network, or the group membership for a single actor, C-Group allows users to focus their analysis on a pair of individuals. While C-Group allows for viewing the addition and deletion of nodes (actors) and edges (relationships) over time, its major contribution is its focus on changing group memberships over time. By doing so, users can investigate the context of temporal group memberships for the pair. C-Group provides users with a flexible interface for defining (and redefining) groups interactively, and supports two novel visual representations of the evolving group memberships. This flexibility gives users alternate views that are appropriate for different network sizes and provides users with different insights into the grouping behavior. We demonstrate the utility of the tool on a scientific publication network.
Collective entity resolution in relational data.
Many databases contain uncertain and imprecise references to real-world entities. The absence of identifiers for the underlying entities often results in a database which contains multiple references to the same entity. This can lead not only to data redundancy, but also inaccuracies in query processing and knowledge extraction. These problems can be alleviated through the use of entity resolution. Entity resolution involves discovering the underlying entities and mapping each database reference to these entities. Traditionally, entities are resolved using pairwise similarity over the attributes of references. However, there is often additional relational information in the data. Specifically, references to different entities may cooccur. In these cases, collective entity resolution, in which entities for cooccurring references are determined jointly rather than independently, can improve entity resolution accuracy. We propose a novel relational clustering algorithm that uses both attribute and relational information for determining the underlying domain entities, and we give an efficient implementation. We investigate the impact that different relational similarity measures have on entity resolution quality. We evaluate our collective entity resolution algorithm on multiple real-world databases. We show that it improves entity resolution performance over both attribute-based baselines and over algorithms that consider relational information but do not resolve entities collectively. In addition, we perform detailed experiments on synthetically generated data to identify data characteristics that favor collective relational resolution over purely attribute-based algorithms.
Probabilistic
Abstract. This paper presents a Bayesian method for constructing probabilistic networks from databases. In particular, we focus on constructing Bayesian belief networks. Potential applications include computer-assisted hypothesis testing, automated scientific discovery, and automated construction of probabilistic expert systems. We extend the basic method to handle missing data and hidden (latent) variables. We show how to perform probabilistic inference by averaging over the inferences of multiple belief networks. Results are presented of a preliminary evaluation of an algorithm for constructing a belief network from a database of cases. Finally, we relate the methods in this paper to previous work, and we discuss open problems.
Entity resolution in geospatial data integration.

Social Capital in Friendship%2DEvent Networks.

Cost%2Dsensitive learning with conditional Markov networks.

Query%2Dtime entity resolution.
Entity resolution is the problem of reconciling database references corresponding to the same real-world entities. Given the abundance of publicly available databases that have unresolved entities, we motivate the problem of query-time entity resolution: quick and accurate resolution for answering queries over such ‘unclean ’ databases at query-time. Since collective entity resolution approaches — where related references are resolved jointly — have been shown to be more accurate than independent attribute-based resolution for off-line entity resolution, we focus on developing new algorithms for collective resolution for answering entity resolution queries at query-time. For this purpose, we first formally show that, for collective resolution, precision and recall for individual entities follow a geometric progression as neighbors at increasing distances are considered. Unfolding this progression leads naturally to a two stage ‘expand and resolve ’ query processing strategy. In this strategy, we first extract the related records for a query using two novel expansion operators, and then resolve the extracted records collectively. We then show how the same strategy can be adapted for query-time entity resolution by identifying and resolving only those database references that are the most helpful for processing the query. We validate our approach on two large real-world publication databases where we show the usefulness of collective resolution and at the same time demonstrate the need for adaptive strategies for query processing. We then show how the same queries can be answered in real-time using our adaptive approach while preserving the gains of collective resolution. In addition to experiments on real datasets, we use synthetically generated data to empirically demonstrate the validity of the performance trends predicted by our analysis of collective entity resolution over a wide range of structural characteristics in the data. 1.
Is there a grand challenge or X%2Dprize for data mining%3F
This panel will discuss possible exciting and motivating Grand Challenge problems for Data Mining, focusing on bioinformatics, multimedia mining, link mining, text mining, and web mining. Categories and Subject Descriptors H.2.8 [Database Applications]: Data mining
A Feature Generation Algorithm for Sequences with Application to Splice%2DSite Prediction.
Abstract. In this paper we present a new approach to feature selection for sequence data. We identify general feature categories and give construction algorithms for each of them. We show how they can be integrated in a system that tightly couples feature construction and feature selection. This integrated process, which we refer to as feature generation, allows us to systematically search a large space of potential features. We demonstrate the effectiveness of our approach for an important component of the gene finding problem, splice-site prediction. We show that predictive models built using our feature generation algorithm achieve a significant improvement in accuracy over existing, state-of-the-art approaches. 1
A Latent Dirichlet Model for Unsupervised Entity Resolution.
Entity resolution has received considerable attention in recent years. Given many references to underlying entities, the goal is to predict which references correspond to the same entity. We show how to extend the Latent Dirichlet Allocation model for this task and propose a probabilistic model for collective entity resolution for relational domains where references are connected to each other. Our approach differs from other recently proposed entity resolution approaches in that it is a) generative, b) does not make pair-wise decisions and c) captures relations between entities through a hidden group variable. We propose a novel sampling algorithm for collective entity resolution which is unsupervised and also takes entity relations into account. Additionally, we do not assume the domain of entities to be known and show how to infer the number of entities from the data. We demonstrate the utility and practicality of our relational entity resolution approach for author resolution in two real-world bibliographic datasets. In addition, we present preliminary results on characterizing conditions under which relational information is useful. 
Name Reference Resolution in Organizational Email Archives.
Online communications provide a rich resource for understanding social networks. Information about the actors, and their dynamic roles and relationships, can be inferred from both the communication content and traffic structure. A key component in the analysis of online communications such as email is the resolution of name references within the body of the message. Name reference resolution relies on the context of the message; both the content of the message and the sender and recipients ’ relationships can help to resolve a reference. Here we investigate a variety of approaches which make use of the email traffic network to disambiguate email name references. The email traffic network serves as a proxy for inferring relationships. These relationships in turn help us infer likely candidates for the name references. Our initial findings suggest that simple temporal models can help us effectively resolve name references. For the class of models proposed, performance is maximized by exploiting long-term traffic statistics to rank candidates. 1
Collective Entity Resolution In Relational Data.
Many databases contain uncertain and imprecise references to real-world entities. The absence of identifiers for the underlying entities often results in a database which contains multiple references to the same entity. This can lead not only to data redundancy, but also inaccuracies in query processing and knowledge extraction. These problems can be alleviated through the use of entity resolution. Entity resolution involves discovering the underlying entities and mapping each database reference to these entities. Traditionally, entities are resolved using pairwise similarity over the attributes of references. However, there is often additional relational information in the data. Specifically, references to different entities may cooccur. In these cases, collective entity resolution, in which entities for cooccurring references are determined jointly rather than independently, can improve entity resolution accuracy. We propose a novel relational clustering algorithm that uses both attribute and relational information for determining the underlying domain entities, and we give an efficient implementation. We investigate the impact that different relational similarity measures have on entity resolution quality. We evaluate our collective entity resolution algorithm on multiple real-world databases. We show that it improves entity resolution performance over both attribute-based baselines and over algorithms that consider relational information but do not resolve entities collectively. In addition, we perform detailed experiments on synthetically generated data to identify data characteristics that favor collective relational resolution over purely attribute-based algorithms.
An Introduction to Probabilistic Graphical Models for Relational Data.

Collective Entity Resolution In Relational Data.
Many databases contain uncertain and imprecise references to real-world entities. The absence of identifiers for the underlying entities often results in a database which contains multiple references to the same entity. This can lead not only to data redundancy, but also inaccuracies in query processing and knowledge extraction. These problems can be alleviated through the use of entity resolution. Entity resolution involves discovering the underlying entities and mapping each database reference to these entities. Traditionally, entities are resolved using pairwise similarity over the attributes of references. However, there is often additional relational information in the data. Specifically, references to different entities may cooccur. In these cases, collective entity resolution, in which entities for cooccurring references are determined jointly rather than independently, can improve entity resolution accuracy. We propose a novel relational clustering algorithm that uses both attribute and relational information for determining the underlying domain entities, and we give an efficient implementation. We investigate the impact that different relational similarity measures have on entity resolution quality. We evaluate our collective entity resolution algorithm on multiple real-world databases. We show that it improves entity resolution performance over both attribute-based baselines and over algorithms that consider relational information but do not resolve entities collectively. In addition, we perform detailed experiments on synthetically generated data to identify data characteristics that favor collective relational resolution over purely attribute-based algorithms.
PRL A probabilistic relational language.

What are the grand challenges for data mining%3F KDD%2D2006 panel report.

Bayesian Network Learning with Abstraction Hierarchies and Context%2DSpecific Independence.

D%2DDupe An Interactive Tool for Entity Resolution in Social Networks.
Visualizing and analyzing social networks is a challenging problem that has been receiving growing attention. An important first step, before analysis can begin, is ensuring that the data is accurate. A common data quality problem is that the data may inadvertently contain several distinct references to the same underlying entity; the process of reconciling these references is called entityresolution. D-Dupe is an interactive tool that combines data mining algorithms for entity resolution with a task-specific network visualization. Users cope with complexity of cleaning large networks by focusing on a small subnetwork containing a potential duplicate pair. The subnetwork highlights relationships in the social network, making the common relationships easy to visually identify. D-Dupe users resolve ambiguities either by merging nodes or by marking them distinct. The entity resolution process is iterative: as pairs of nodes are resolved, additional duplicates may be revealed; therefore, resolution decisions are often chained together. We give examples of how users can flexibly apply sequences of actions to produce a high quality entity resolution result. We illustrate and evaluate the benefits of D-Dupe on three bibliographic collections. Two of the datasets had already been cleaned, and therefore should not have contained duplicates; despite this fact, many duplicates were rapidly identified using D-Dupe’s unique combination of entity resolution algorithms within a task-specific visual interface.
Pruning Social Networks Using Structural Properties and Descriptive Attributes.
Scale is often an issue with understanding and making sense of large social networks. Here we investigate methods for pruning social networks by determining the most relevant relationships. We measure importance in terms of predictive accuracy on a set of target attributes of the social network. Our goal is to create a pruned network that models only the most informative affiliations and relationships. We present methods for pruning networks based on both structural properties and descriptive attributes demonstrate it on a network of NASDAQ and NYSE businesses and on a bibliographic network. 1
Tutorial on Statistical Relational Learning.

05051 Abstracts Collection %2D Probabilistic

05051 Executive Summary %2D Probabilistic

Introduction to the special issue on link mining.

Link mining a survey.

Unsupervised Sense Disambiguation Using Bilingual Probabilistic Models.
We describe two probabilistic models for unsupervised word-sense disambiguation using parallel corpora. The first model, which we call the Sense model, builds on the work of Diab and Resnik (2002) that uses both parallel text and a sense inventory for the target language, and recasts their approach in a probabilistic framework. The second model, which we call the Concept model, is a hierarchical model that uses a concept latent variable to relate different language specific sense labels. We show that both models improve performance on the word sense disambiguation task over previous unsupervised approaches, with the Concept model showing the largest improvement. Furthermore, in learning the Concept model, as a by-product, we learn a sense inventory for the parallel language. 1
Iterative record linkage for cleaning and integration.

Using the Structure of Web Sites for Automatic Segmentation of Tables.
Many Web sites, especially those that dynamically generate HTML pages to display the results of a user’s query, present information in the form of list or tables. Current tools that allow applications to programmatically extract this information rely heavily on user input, often in the form of labeled extracted records. The sheer size and rate of growth of the Web make any solution that relies primarily on user input is infeasible in the long term. Fortunately, many Web sites contain much explicit and implicit structure, both in layout and content, that we can exploit for the purpose of information extraction. This paper describes an approach to automatic extraction and segmentation of records from Web tables. Automatic methods do not require any user input, but rely solely on the layout and content of the Web source. Our approach relies on the common structure of many Web sites, which present information as a list or a table, with a link in each entry leading to a detail page containing additional information about that item. We describe two algorithms that use redundancies in the content of table and detail pages to aid in information extraction. The first algorithm encodes additional information provided by detail pages as constraints and finds the segmentation by solving a constraint satisfaction problem. The second algorithm uses probabilistic inference to find the record segmentation. We show how each approach can exploit the web site structure in a general, domain-independent manner, and we demonstrate the effectiveness of each algorithm on a set of twelve Web sites. 1.
Understanding tuberculosis epidemiology using structured statistical models.


A workshop report mining for and from the Semantic Web at KDD 2004.

Proceedings of the Ninth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining
The problem of identifying approximately duplicate records in databases is an essential step for data cleaning and data integration processes. Most existing approaches have relied on generic or manually tuned distance metrics for estimating the similarity of potential duplicates. In this paper, we present a framework for improving duplicate detection using trainable measures of textual similarity. We propose to employ learnable text distance functions for each database field, and show that such measures are capable of adapting to the specific notion of similarity that is appropriate for the field&#039;s domain. We present two learnable text similarity measures suitable for this task: an extended variant of learnable string edit distance, and a novel vector-space based measure that employs a Support Vector Machine (SVM) for training. Experimental results on a range of datasets show that our framework can improve duplicate detection accuracy over traditional techniques.
PXML A Probabilistic Semistructured Data Model and Algebra.
ehung,getoor,vs£ Despite the recent proliferation of work on semistructured data models, there has been little work to date on supporting uncertainty in these models. In this paper, we propose a model for probabilistic semistructured data (PSD). The advantage of our approach is that it supports a flexible representation that allows the specification of a wide class of distributions over semistructured instances. We provide two semantics for the model and show that the semantics are probabilistically coherent. Next, we develop an extension of the relational algebra to handle probabilistic semistructured data and describe efficient algorithms for answering queries that use this algebra. Finally, we present experimental results showing the efficiency of our algorithms. 1

Link%2Dbased Classification.
There has been a surge of interest in learning using a mix of labeled and unlabeled data. General approaches include semi-supervised learning and tranductive inference. In this paper we look at some of the unique ways in which unlabeled data can improve performance when doing link-based classification, the classification of objects making use of both object descriptions and the links between objects. 1.
Learning Structure From Statistical Models.

Link mining a new data mining challenge.

Learning Probabilistic Models of Link Structure.
Most real-world data is heterogeneous and richly interconnected. Examples include the Web, hypertext,  bibliometric data and social networks. In contrast, most statistical learning methods work with &#034;flat&#034; data  representations, forcing us to convert our data into a form that loses much of the link structure. The recently  introduced framework of probabilistic relational models (PRMs) embraces the object-relational nature  of structured data by capturing probabilistic interactions between attributes of related entities. In this paper,  we extend this framework by modeling interactions between the attributes and the link structure itself. An  advantage of our approach is a unified generarive model for both content and relational structure. We propose  two mechanisms for representing a probabilistic distribution over link structures: reference uncertainty and  existence uncertainty. We describe the appropriate conditions for using each model and present learning algorithms  for each. We present experimental results showing that the learned models can be used to predict link  structure and, moreover, the observed link structure can be used to provide better predictions for the attributes  in the model.
Learning Probabilistic Models of Relational Structure.
Most real-world data is stored in relational form. In contrast, most statistical learning methods work with “flat ” data representations, forcing us to convert our data into a form that loses much of the relational structure. The recently introduced framework of probabilistic relational models (PRMs) allows us to represent probabilistic models over multiple entities that utilize the relations between them. In this paper, we propose the use of probabilistic models not only for the attributes in a relational model, but for the relational structure itself. We propose two mechanisms for modeling structural uncertainty: reference uncertainty and existence uncertainty. We describe the appropriate conditions for using each model and present learning algorithms for each. We present experimental results showing that the learned models can be used to predict relational structure and, moreover, the observed relational structure can be used to provide better predictions for the attributes in the model. 1.
Selectivity Estimation using Probabilistic Models.
Estimating the result size of complex queries that involve selection on multiple attributes and the join of several relations is a difficult but fundamental task in database query processing. It arises in cost-based query optimization, query profiling, and approximate query answering. In this paper, we show how probabilistic graphical models can be effectively used for this task as an accurate and compact approximation of the joint frequency distribution of multiple attributes across multiple relations. Probabilistic Relational Models (PRMs) are a recent development that extends graphical statistical models such as Bayesian Networks to relational domains. They represent the statistical dependencies between attributes within a table, and between attributes across foreign-key joins. We provide an efficient algorithm for constructing a PRM from a database, and show how a PRM can be used to compute selectivity estimates for a broad class of queries. One of the major contributions of this work is a unified framework for the estimation of queries involving both select and foreign-key join operations. Furthermore, our approach is not limited to answering a small set of predetermined queries; a single model can be used to effectively estimate the sizes of a wide collection of potential queries across multiple tables. We present results for our approach on several real-world databases. For both single-table multi-attribute queries and a general class of select-join queries, our approach produces more accurate estimates than standard approaches to selectivity estimation, using comparable space and time.
AAAI 2000 Workshop Reports.

Using Feature Hierarchies in Bayesian Network Learning.
In recent years, researchers in statistics and the UAI community have developed an impressive body of theory and algorithmic machinery for learning Bayesian networks from data. Learned Bayesian networks can be used for pattern discovery, prediction, diagnosis, and density estimation tasks. Early pioneering work in this area includes [5, 9, 10, 13]. The algorithm that has emerged as the current most popular approach is a simple greedy hill-climbing algorithm that searches the space of candidate structures, guided by a network scoring function (either Bayesian or Minimum Description Length (MDL)-based). The search begins with an initial candidate network (typically the empty network, which has no edges), and then considers making small local changes such as ...
Learning Probabilistic Relational Models.
A large portion of real-world data is stored in commercial relational database systems. In contrast, most statistical learning methods work only with &amp;quot;flat &amp;quot; data representations. Thus, to apply these methods, we are forced to convert our data into a flat form, thereby losing much of the relational structure present in our database. This paper builds on the recent work on probabilistic relational models (PRMs), and describes how to learn them from databases. PRMs allow the properties of an object to depend probabilistically both on other properties of that object and on properties of related objects. Although PRMs are significantly more expressive than standard models, such as Bayesian networks, we show how to extend well-known statistical methods for learning Bayesian networks to learn these models. We describe both parameter estimation and structure learning — the automatic induction of the dependency structure in a model. Moreover, we show how the learning procedure can exploit standard database retrieval techniques for efficient learning from large datasets. We present experimental results on both real and synthetic relational databases. 1
Learning Probabilistic Relational Models.
A large portion of real-world data is stored in commercial relational database systems. In contrast, most statistical learning methods work only with &amp;quot;flat &amp;quot; data representations. Thus, to apply these methods, we are forced to convert our data into a flat form, thereby losing much of the relational structure present in our database. This paper builds on the recent work on probabilistic relational models (PRMs), and describes how to learn them from databases. PRMs allow the properties of an object to depend probabilistically both on other properties of that object and on properties of related objects. Although PRMs are significantly more expressive than standard models, such as Bayesian networks, we show how to extend well-known statistical methods for learning Bayesian networks to learn these models. We describe both parameter estimation and structure learning — the automatic induction of the dependency structure in a model. Moreover, we show how the learning procedure can exploit standard database retrieval techniques for efficient learning from large datasets. We present experimental results on both real and synthetic relational databases. 1
Utility Elicitation as a Classification Problem.
We investigate the application of classification techniques  to utility elicitation. In a decision problem, two  sets of parameters must generally be elicited: the probabilities  and the utilities. While the prior and conditional  probabilities in the model do not change from  user to user, the utility models do. Thus it is necessary  to elicit a utility model separately for each new user. Elicitation is 
Effective Redundant Constraints for Online Scheduling.
The use of heuristics as a means to improve constraint solver performance has been researched widely. However, most work has been on problem-independent heuristics (e.g., variable and value ordering), and has focused on offline problems (e.g., one-shot constraint satisfaction). In this paper, we present an online scheduling problem for which we are developing a real-time scheduling algorithm. While we can and do use generic heuristics in the scheduler, here we focus on the use of domain-specific redundant constraints to effectively approximate optimal offline solutions. We present a taxonomy of redundant domain constraints, and examine their impact on the effectiveness of the scheduler. We also describe several techniques for generating redundant constraints, which can be applied to a large class of job shop scheduling problems.
Scope and Abstraction Two Criteria for Localized Planning.
ion: Two Criteria for Localized Planning  Amy L. Lansky Lise C. Getoor  Recom Technologies/NASA Ames Research Center Artificial Intelligence Research Branch MS 269-2, Moffett Field, CA 94035-1000  LANSKY@PTOLEMY.ARC.NASA.GOV GETOOR@PTOLEMY.ARC.NASA.GOV Abstract  Localization is a general-purpose representational technique for partitioning a problem into subproblems. A localized problem-solver searches several smaller search spaces, one for each subproblem. Unlike most methods of partitioning, however, localization allows for subproblems that overlap -- i.e. multiple search spaces may be involved in constructing shared pieces of the overall plan. In this paper we focus on two criteria for forming localizations: scope  and abstraction. We describe a method for automatically generating such localizations and provide empirical results that contrast their use in an office-building construction domain.  1 Introduction  Over the years, many researchers have focused on the use of abstraction t...
