Unsupervised Entity Resolution on Multi-type Graphs

Zhu, Linhong; Ghasemi-Gol, Majid; Szekely, Pedro; Galstyan, Aram; Knoblock, Craig A.

doi:10.1007/978-3-319-46523-4_39

Cited by 29 publications

(30 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Other Entity Resolution Systems. In order to set the GenLink results into context, we also ran the WDC Gold Standard experiments with EAGLE [36], a supervised matching system that also employs genetic programming, 3 , FEBRL [9] 4 an entity resolution system that internally employs an SVM, and CoSum-P [46], an unsupervised system that treats entity resolution as a graph summarization problem. We pre-compute attribute similarities for CoSum-P as described in [46].…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Learning expressive linkage rules from sparse data

Petrovski

Bizer

2020

View full text Add to dashboard Cite

A central problem in the context of the Web of Data, as well as in data integration in general is to identify entities in different data sources that describe the same real-world object. There exists a large body of research on entity resolution. Interestingly, most of the existing research focuses on entity resolution on dense data, meaning data that does not contain too many missing values. This paper sets a different focus and explores learning expressive linkage rules from as well as applying these rules to sparse web data, i.e. data exhibiting a large amount of missing values. Such data is a common challenge in various application domains including e-commerce, online hotel booking, or online recruiting. We propose and compare three entity resolution methods that employ genetic programming to learn expressive linkage rules from sparse data. First, we introduce the GenLinkGL algorithm which learns groups of matching rules and applies specific rules out of these groups depending on which values are missing from a pair of records. Next, we propose GenLinkSA, which employs selective aggregation operators within rules. These operators exclude misleading similarity scores (which result from missing values) from the aggregations, but on the other hand also penalize the uncertainty that results from missing values. Finally, we introduce GenLinkComb, a method which combines the central ideas of the previous two into one integrated method. We evaluate all methods using six benchmark datasets: three of them are e-commerce product datasets, the other datasets describe restaurants, movies, and drugs. We show improvements of up to 16% F-measure compared to handwritten rules, on average 12% F-measure improvement compared to the original GenLink algorithm, 15% compared to EAGLE, 8% compared to FEBRL, and 5% compared to CoSum-P.

show abstract

Section: Methodsmentioning

confidence: 99%

“…Moreover, we compare results from: (i) handwritten matching rules;, (ii) the GenLink algorithm, (iii) GenLinkGL, (iv) Gen-LinkSA and (v) GenLinkComb. Additionally, we compare to three state-of-the-art matching systems for this dataset: (i) EAGLE [36], (ii) FEBRL [23] and (iii) CoSum-P [46] as explained above.…”

Section: Methodsmentioning

confidence: 99%

Learning expressive linkage rules from sparse data

Petrovski

Bizer

2020

View full text Add to dashboard Cite

show abstract

“…To run baselines that do not explicitly support heterogeneous graphs, we align nodes of the input graph according to their object types and re-order the IDs to form the homogeneous representation. In node aggregation, CoSum [33] ran out of memory due to the computation of pairwise node similarity. We use Louvain [2] as an alternative that scales to large graphs and forms the basis of many node aggregation methods.…”

Section: Datamentioning

confidence: 99%

Latent Network Summarization

Jin

Rossi

Koh

et al. 2019

Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery &Amp; Data Mining

View full text Add to dashboard Cite

Motivated by the computational and storage challenges that dense embeddings pose, we introduce the problem of latent network summarization that aims to learn a compact, latent representation of the graph structure with dimensionality that is independent of the input graph size (i.e., #nodes and #edges), while retaining the ability to derive node representations on the fly. We propose Multi-LENS, an inductive multi-level latent network summarization approach that leverages a set of relational operators and relational functions (compositions of operators) to capture the structure of egonets and higher-order subgraphs, respectively. The structure is stored in lowrank, size-independent structural feature matrices, which along with the relational functions comprise our latent network summary. Multi-LENS is general and naturally supports both homogeneous and heterogeneous graphs with or without directionality, weights, attributes or labels. Extensive experiments on real graphs show 3.5 − 34.3% improvement in AUC for link prediction, while requiring 80 − 2152× less output storage space than baseline embedding methods on large datasets. As application areas, we show the effectiveness of Multi-LENS in detecting anomalies and events in the Enron email communication graph and Twitter co-mention graph.

show abstract

“…Two papers have won Best Paper awards [9], [10]. More broadly, our experience has spurred us to write a graduate-level textbook on knowledge graphs, which will be published by MIT Press later this year.…”

Section: Research Outputsmentioning

confidence: 99%

“…DIG has yielded over 15 peer-reviewed publications over the course of three years, some examples being [9], [10], [6], [8], [5], [2], [4], [3], [1] with several more under review, and at least one dataset resource [7]. Two papers have won Best Paper awards [9], [10].…”

Section: Research Outputsmentioning

confidence: 99%

Domain-specific Insight Graphs (DIG)

Szekely¹,

Kejriwal²

2018

Companion of the the Web Conference 2018 on the Web Conference 2018 - WWW '18

Self Cite

View full text Add to dashboard Cite

The DARPA Memex program was established with the goal of funding research into building domain-specific search systems that integrated state-of-the-art focused crawling ('domain discovery') information extraction and semantic search, and that could be used by users and domain experts with no programming or technical experience. Domain-specific Insight Graphs (DIG) was proposed and funded under Memex and has led to an end-to-end search system currently being used by over 200 law enforcement for combating human trafficking, by investigators from the Securities and Exchange Commission (SEC) in the US for investigating securities fraud, and for numerous other domains of a difficult, socially consequential (e.g., investigative) and unusual nature.

show abstract

Unsupervised Entity Resolution on Multi-type Graphs

Cited by 29 publications

References 31 publications

Learning expressive linkage rules from sparse data

Learning expressive linkage rules from sparse data

Latent Network Summarization

Domain-specific Insight Graphs (DIG)

Contact Info

Product

Resources

About