T2LD : Interpreting and Representing Tables as Linked Data?

Abstract

We describe a framework and prototype system for interpreting tables and extracting entities and relations from them, and producing a linked data representation of the table's contents. This can be used to annotate the table or to add new facts to the linked data collection.

Introduction

Existing systems for extracting knowledge from tables [ 5 ] require human intervention and do not focus on a complete interpretation of the table, nor integrating the table with linked open data cloud. This poster paper focuses on an automatic framework for generating an linked RDF which can be integrated into the LOD cloud. The eventual goal of this work is to enrich the LOD cloud by learning new facts and knowledge from tables and publishing it on the Semantic Web.

To develop an overall interpretation of a table, we assign every column header city state mayor population a class label from an appropriate ontol- PBhiallatdi meloprheia MPAD MS..DNiuxtotner 1654000000000 ogy, e.g., the column with header City Washington DC A.Fenty 595000 is assigned a class label dbpedia-owl:City New York NY M.Bloomberg 8400000 from the DBpedia ontology. For the ta- Boston MA T.Menino 610000 ble in Figure 1, we link \Baltimore" to Fig. 1: In simple tables column headers suggests dbpedia:Baltimore. Numbers can be map- the type of data stored in columns and cell values denote instances of that type. ped as values of properties which can be associated with entities in the table.

We also identify the relations implicit between columns, e.g., that dbpediaowl:largestCity seems to hold between the entities denoted by cell values in the rst two columns (i.e., city and state). Finally this information is represented in a N3 serialization of RDF. 2

T2LD Framework

Given an table as input, the T2LD framework [ 6 ] begins with the process of assigning a class label to every column in the table. For all the cell values in every column of the table, the algorithm for assigning class labels (see Algorithm 1 in [ 3 ]) submits a complex query to the Wikitology knowledge base to determine the type of each cell value in the column. Each class label from the set of possible class labels obtained from query results is scored. The class label with the highest score is chosen as the class label to be associated with the column. We predict class labels from four vocabularies - DBpedia Ontology, Freebase, WordNet, and Yago.

Using the class labels as additional evidence, for every MAP columns table cell, the algorithm for linking table cell to entities (see m = 1 11.53% Algorithm 2 in [ 3 ] for detailed algorithm), re-queries the KB. 0 <mm= <0 1 1699..2243%% For every table cell, the KB returns the top N possible enti- Recall columns ties. For each of the top N entities, the algorithm generates r = 1 46.15% a feature vector consisting of the entity's KB score, entity's 0 <r =r <0 1 3149..6214%% Wikipedia page length, entity's page rank, the Levenshtein distance between the entity and the string in the query and Fig. 2: The percentthe Dice score between the entity and the string. The set avagreiooufscMolAumPnasndwirtehfeature vectors for each table cell are ranked using a SVM- call scores. Rank classi er. To the highest rank feature vector from SVM rank, two more features are added - the SVM rank score of the feature vector and the di erence \City"@en is rdfs:label of dbpedia-owl:City . \State"@en is rdfs:label of dbpedia-owl:AdminstrativeRegion . \Baltimore"@en is rdfs:label of dbpedia:Baltimore . dbpedia:Baltimore a dbpedia-owl:City . \MD"@en is rdfs:label of dbpedia:Maryland . dbpedia:Maryland a dbpedia-owl:AdministrativeRegion . dbpprop:LargestCity rdfs:domain dbpedia-owl:AdminstrativeRegion . dbpprop:LargestCity rdfs:range dbpedia-owl:City . in SVM-Rank scores between the top two feature vectors. Based on this new feature vector, a second SVM classi er decides whether to link the table cell to this top ranked entity or not. If the evidence is not strong enough, it is likely that the table cell is a new entity not present in the KB; this step is useful in discovery of new entities in a given table. If the evidence is strong enough, the table cell is linked to the top ranked entity returned by SVM-Rank.

We also present a preliminary approach for identifying relations between table columns (see Algorithm 3 in [ 3 ]). The algorithm generates a set of candidate relations from the relations that exist between the strings in each row of the two columns. Each candidate relation is scored and the relation with the highest score is selected to represent relation between the two columns. We have also developed a preliminary template in N3 (see Figure 3), which is a compact and human readable serialization of RDF for representing tables as LOD. 3

Evaluation and Conclusion

Our implemented prototype was evaluated against 15 tables obtained from Google Squared, Wikipedia and from a collection of tables extracted from the Web. Excluding the columns with numbers, the 15 tables have 52 columns and 611 entities for evaluation of our algorithms. We used a subset of 23 columns for evaluation of relation identifcation between columns.

In the rst evaluation of the algorithm for assigning class labels to columns, we compared the ranked list of possible class labels generated by the system against the list of possible class labels ranked by the evaluators. As shown in Figure 2 for 80.76 % of the columns the Mean Average Precision (MAP) between the system and evaluators list is greater than 0 which indicates that there was at least one relevant label in the top three of the system ranked list. Also seen in Figure 2, for 75 % of the columns, the recall of the algorithm was greater than or equal to 0.6. We also assessed whether our predicted class labels were reasonable based on the judgment of human subjects (see [ 3 ]). 76.92 % of the class labels predicted were considered correct by the evaluators. The accuracy in each of the four categories is shown in Figure 4. 66.12 % of the table cell strings were correctly linked by our algorithm for linking table cells. The breakdown of accuracy based on the categories is shown in Figure 4. Our dataset had 24 new entities and our algorithm was able to correctly predict for all the 24 entities as new entities not present in the KB. We did not get encouraging results for relationship identi cation with an accuracy of 25 % (see [ 3 ] for details).

Our existing system performs reasonably well in selecting appropriate types for columns and linking cell values to LOD entities. We have preliminary results for identifying and encoding the relationships implicit in the columns as well. Our current work is focused on improving relationship discovery and generating new facts and knowledge from tables that contain entities not present in the LOD knowledge bases.

References

1. Finin , T. , Syed , Z. : Creating and Exploiting a Web of Semantic Data . In: Proc. 2nd Int. Conf. on Agents and Arti cial Intelligence , Springer ( 2010 )

2. Cafarella , M.J. , Halevy , A.Y. , Wang , Z.D. , Wu , E. , Zhang, Y.: Webtables: exploring the power of tables on the web . PVLDB 1 ( 2008 ) 538 { 549

3. Mulwad , V.: T2LD - An automatic framework for extracting, interpreting and representing tables as Linked Data . Master's thesis , U. of Maryalnd, Baltimore County ( 2010 )

4. Sahoo , S.S. , Halb , W. , Hellmann , S. , Idehen , K. , Thibodeau Jr, T. , Auer , S. , Sequeda , J. , Ezzat , A. : A survey of current approaches for mapping of relational databases to rdf . Technical report, W3C ( 2009 )

5. Han , L. , Finin , T. , Parr , C. , Sachs , J. , Joshi , A. : RDF123: from Spreadsheets to RDF . In: Seventh International Semantic Web Conference, Springer ( 2008 )

6. Syed , Z. , Finin , T. , Mulwad , V. , Joshi , A. : Exploiting a Web of Semantic Data for Interpreting Tables . In: Proceedings of the Second Web Science Conference . ( 2010 )

7. Mulwad , V. , Finin , T. , Syed , Z. , Joshi , A. : Using linked data to interpret tables . In: Proc. First Int. Workshop on Consuming Linked Data . ( 2010 )