No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Towards More Accurate Statistical Profiling of Deployed schema.org Microdata
Abstract
Being promoted by major search engines such as Google, Yahoo!, Bing, and Yandex, Microdata embedded in web pages, especially using schema.org, has become one of the most important markup languages for the Web. However, deployed Microdata is very often not free from errors, which makes it difficult to estimate the data volume and create an accurate data profile. In addition, as the usage of global identifiers is not common, the real number of entities described by this format in the Web is hard to assess. In this article, we discuss how the subsequent application of data cleaning steps, such as duplicate detection and correction of common schema-based errors, leads to a more realistic view on the data, step by step. The cleaning steps applied include both heuristics for fixing errors as well as means to perform duplicate detection and duplicate elimination. Using the Web Data Commons Microdata corpus, we show that applying such quality improvement methods can essentially change the statistical profile of the dataset and lead to different estimates of both the number of entities as well as the class distribution within the data.
... Through its wide availability, markup lends itself as a diverse source of input data for KBA. In particular when attempting to complement information about long-tail attributes and entities, the diversity and scale of markup provide opportunities for enriching existing knowledge bases and graphs [23]. ...
... 7 The mappings are online at: http://l3s.de/ yu/knowmore/ Tail entities/types as well as time-dependent properties which require frequent updates, such as the award of a book, are prevalent in markup data [23], yet tend to be underrepresented in structured KBs. Hence, markup data lends itself as data source for the KBA task. ...
... For instance, taxID can be considered one of the pseudo-key property for instances of type Person, and isbn for instances of type Book. However, as studied by Meusel et al. [23], resolving coreferences simply through pseudo-key properties does not produce sufficient results when applied on sparsely described and heterogeneous entity descriptions obtained from Web markup. ...
Knowledge bases are in widespread use for aiding tasks such as information extraction and information retrieval, for example in Web search. However, knowledge bases are known to be inherently incomplete, where in particular tail entities and properties are under-represented. As a complimentary data source, embedded entity markup based on Microdata, RDFa, and Microformats have become prevalent on the Web and constitute an unprecedented source of data with significant potential to aid the task of knowledge base augmentation (KBA). RDF statements extracted from markup are fundamentally different from traditional knowledge graphs: entity descriptions are flat, facts are highly redundant and of varied quality, and, explicit links are missing despite a vast amount of coreferences. Therefore, data fusion is required in order to facilitate the use of markup data for KBA. We present a novel data fusion approach which addresses these issues through a combination of entity matching and fusion techniques geared towards the specific challenges associated with Web markup. To ensure precise and non-redundant results, we follow a supervised learning approach based on a set of features considering aspects such as quality and relevance of entities, facts and their sources. We perform a thorough evaluation on a subset of the Web Data Commons dataset and show significant potential for augmenting existing knowledge bases. A comparison with existing data fusion baselines demonstrates superior performance of our approach when applied to Web markup data.
... complemented by a long tail of less frequent statements. Moreover, data extracted from markup contains a wide variety of errors [17], ranging from typos to the frequent misuse of vocabulary terms [15]. Hence, individual markup extracted from a particular Web document or crawl usually contains very limited or unreliable information about a particular entity. ...
... e s:name of a subject is used as input for DBpedia Spotlight [5] to obtain candidate entities from DBpedia (dbp). If the markup is annotated in one of the 12 languages supported by Spotlight 17 , the corresponding Spotlight model is used. For all other cases we employ the English Spotlight model. ...
Embedded markup of Web pages has seen widespread adoption throughout the past years driven by standards such as RDFa and Microdata and initiatives such as schema.org, where recent studies show an adoption by 39% of all Web pages already in 2016. While this constitutes an important information source for tasks such as Web search, Web page classification or knowledge graph augmentation, individual markup nodes are usually sparsely described and often lack essential information. For instance, from 26 million nodes describing events within the Common Crawl in 2016, 59% of nodes provide less than six statements and only 257,000 nodes (0.96%) are typed with more specific event subtypes. Nevertheless, given the scale and diversity of Web markup data, nodes that provide missing information can be obtained from the Web in large quantities, in particular for categorical properties. Such data constitutes potential training data for inferring missing information to significantly augment sparsely described nodes. In this work, we introduce a supervised approach for inferring missing categorical properties in Web markup. Our experiments, conducted on properties of events and movies, show a performance of 79% and 83% F1 score correspondingly, significantly outperforming existing baselines.
... Other examples of work that are related to schema.org analysis but that are too extensive to describe here include (Nam and Kejriwal 2018), (Mihindukulasooriya et al. 2017), (Beek et al. 2014), (Abedjan et al. 2014), (Abedjan, Lorey, and Naumann 2012) and (Meusel, Ritze, and Paulheim 2016). ...
Schema.org has experienced high growth in recent years. Structured descriptions of products embedded in HTML pages are now not uncommon, especially on e-commerce websites. The Web Data Commons (WDC) project has extracted schema.org data at scale from webpages in the Common Crawl and made it available as an RDF `knowledge graph' at scale. The portion of this data that specifically describes products offers a golden opportunity for researchers and small companies to leverage it for analytics and downstream applications. Yet, because of the broad and expansive scope of this data, it is not evident whether the data is usable in its raw form. In this paper, we do a detailed empirical study on the product-specific schema.org data made available by WDC. Rather than simple analysis, the goal of our study is to devise an empirically grounded set of best practices for using and consuming WDC product-specific schema.org data. Our studies reveal five best practices, each of which is justified by experimental data and analysis.
... Other examples of work that are related to schema.org analysis but that are too extensive to describe here include [12], [10], [18], [3], [1], [2] and [9]. ...
Schema.org has experienced high growth in recent years. Structured descriptions of products embedded in HTML pages are now not uncommon, especially on e-commerce websites. The Web Data Commons (WDC) project has extracted schema.org data at scale from webpages in the Common Crawl and made it available as an RDF `knowledge graph' at scale. The portion of this data that specifically describes products offers a golden opportunity for researchers and small companies to leverage it for analytics and downstream applications. Yet, because of the broad and expansive scope of this data, it is not evident whether the data is usable in its raw form. In this paper, we do a detailed empirical study on the product-specific schema.org data made available by WDC. Rather than simple analysis, the goal of our study is to devise an empirically grounded set of best practices for using and consuming WDC product-specific schema.org data. Our studies reveal six best practices, each of which is justified by experimental data and analysis.
... However, some Web crawls have been made openly available, such as the Common Crawl corpus which contains "petabytes of data collected over the last 7 years" 10 . Indeed the project has already been used to collect and analyse the availability (and quality) of structured data on the Web, e.g. in the Web Data Commons Project [50,51]. Open Data portals are collections or catalogs that index metadata and link to actual data resources which have become popular over the past few years through various Open Government Data Initiatives, but also in the private sector. ...
In this lecture we will discuss and introduce challenges of integrating openly available Web data and how to solve them. Firstly, while we will address this topic from the viewpoint of Semantic Web research, not all data is readily available as RDF or Linked Data, so we will give an introduction to different data formats prevalent on the Web, namely, standard formats for publishing and exchanging tabular, tree-shaped, and graph data. Secondly, not all Open Data is really completely open, so we will discuss and address issues around licences, terms of usage associated with Open Data, as well as documentation of data provenance. Thirdly, we will discuss issues connected with (meta-)data quality issues associated with Open Data on the Web and how Semantic Web techniques and vocabularies can be used to describe and remedy them. Fourth, we will address issues about searchability and integration of Open Data and discuss in how far semantic search can help to overcome these. We close with briefly summarizing further issues not covered explicitly herein, such as multi-linguality, temporal aspects (archiving, evolution, temporal querying), as well as how/whether OWL and RDFS reasoning on top of integrated open data could be help.
The web of data is growing constantly, both in terms of size and impact. A potential data publisher needs to dispose with recapitulative information on the datasets available on the web, so that she can easily identify where to look for the resources to which her data relates. This information will help discover candidate datasets for interlinking. In that context, we investigate the problem of dataset profiling. We define a dataset profile as a set of characteristics, both semantic and statistical, that allow to describe in the best possible way a dataset by taking into account the multiplicity of domains and vocabularies on the web of data.
It is widely accepted that proper data publishing is difficult. The majority of Linked Open Data (LOD) does not meet even a core set of data publishing guidelines. Moreover, datasets that are clean at creation, can get stains over time. As a result, the LOD cloud now contains a high level of dirty data that is difficult for humans to clean and for machines to process.
Existing solutions for cleaning data (standards, guidelines, tools) are targeted towards human data creators, who can (and do) choose not to use them. This paper presents the LOD Laundromat which removes stains from data without any human intervention. This fully automated approach is able to make very large amounts of LOD more easily available for further processing right now.
LOD Laundromat is not a new dataset, but rather a uniform point of entry to a collection of cleaned siblings of existing datasets. It provides researchers and application developers a wealth of data that is guaranteed to conform to a specified set of best practices, thereby greatly improving the chance of data actually being (re)used.
The development and standardization of Semantic Web technologies has resulted in an unprecedented volume of data being published on the Web as Linked Data (LD). However, we observe widely varying data quality ranging from extensively curated datasets to crowdsourced and extracted data of relatively low quality. In this article, we present the results of a systematic review of approaches for assessing the quality of LD. We gather existing approaches and analyze them qualitatively. In particular, we unify and formalize commonly used terminologies across papers related to data quality and provide a comprehensive list of 18 quality dimensions and 69 metrics. Additionally, we qualitatively analyze the 30 core approaches and 12 tools using a set of attributes. The aim of this article is to provide researchers and data curators a comprehensive understanding of existing work, thereby encouraging further experimentation and development of new approaches focused towards data quality, specifically for LD.
The datasets that are part of the Linking Open Data cloud diagramm (LOD cloud) are classified into the following topical categories: media, government, publications, life sciences, geographic, social networking, user-generated content, and cross-domain. The topical categories were manually assigned to the datasets. In this paper, we investigate to which extent the topical classification of new LOD datasets can be automated using machine learning techniques and the existing annotations as supervision. We conducted experiments with different classification techniques and different feature sets. The best classification technique/feature set combination reaches an accuracy of 81:62% on the task of assigning one out of the eight classes to a given LOD dataset. A deeper inspection of the classification errors reveals problems with the manual classification of datasets in the current LOD cloud.
In the recent years, the publication of structured data inside HTML content of Web sites has become a mainstream feature of commercial Web sites. In particular, e-commerce sites have started to add RDFa or Microdata markup based on schema.org and GoodRelations vocabularies. For many potential usages of this huge body of data, we need to crawl the sites and extract the data from the markup. Unfortunately, a lot of markup resides in very deep branches of the sites, namely in the product detail pages. Such pages are difficult to crawl because of their sheer number and because they often lack links pointing to them. In this paper, we present a small-sized experiment where we compare the Web pages from a popular Web crawler, Common Crawl, with the URLs in sitemap files of respective Web sites. We show that Common Crawl lacks a major share of the product detail pages that hold a significant part of the data, and that an approach as simple as a sitemap crawl yields much more product pages. Based on our findings, we conclude that a rethinking of state-of-the-art crawling strategies is necessary in order to cater for e-commerce scenarios.
Entity augmentation is a query type in which, given a set of entities and a large corpus of possible data sources, the values of a missing attribute are to be retrieved. State of the art methods return a single result that, to cover all queried entities, is fused from a potentially large set of data sources. We argue that queries on large corpora of heterogeneous sources using information retrieval and automatic schema matching methods can not easily return a single result that the user can trust, especially if the result is composed from a large number of sources that user has to verify manually. We therefore propose to process these queries in a Top-k fashion, in which the system produces multiple minimal consistent solutions from which the user can choose to resolve the uncertainty of the data sources and methods used. In this paper, we introduce and formalize the problem of consistent, multi-solution set covering, and present algorithms based on a greedy and a genetic optimization approach. We then apply these algorithms to Web table-based entity augmentation. The publication further includes a Web table corpus with 100M tables, and a Web table retrieval and matching system in which these algorithms are implemented. Our experiments show that the consistency and minimality of the augmentation results can be improved using our set covering approach, without loss of precision or coverage and while producing multiple alternative query results.
Before reaping the benefits of open data to add value to an organizations internal data, such new, external datasets must be analyzed and understood already at the basic level of data types, constraints, value patterns etc. Such data profiling, already difficult for large relational data sources, is even more challenging for RDF datasets, the preferred data model for linked open data. We present ProLod++, a novel tool for various profiling and mining tasks to understand and ultimately improve open RDF data. ProLod++ comprises various traditional data profiling tasks, adapted to the RDF data model. In addition, it features many specific profiling results for open data, such as schema discovery for user-generated attributes, association rule discovery to uncover synonymous predicates, and uniqueness discovery along ontology hierarchies. ProLod++ is highly efficient, allowing interactive profiling for users interested in exploring the properties and structure of yet unknown datasets.
More and more websites embed structured data describing for in-stance products, people, organizations, places, events, resumes, and cooking recipes into their HTML pages using encoding standards such as Microformats, Microdatas and RDFa. The Web Data Com-mons project extracts all Microformat, Microdata and RDFa data from the Common Crawl web corpus, the largest and most up-to-data web corpus that is currently available to the public, and provides the extracted data for download in the form of RDF-quads. In this paper, we give an overview of the project and present statistics about the popularity of the different encoding standards as well as the kinds of data that are published using each format.
The increasing adoption of Linked Data principles has led to an abundance of datasets on the Web. However, take-up and reuse is hindered by the lack of descriptive information about the nature of the data, such as their topic coverage, dynamics or evolution. To address this issue, we propose an approach for creating linked dataset profiles. A profile consists of structured dataset metadata describing topics and their relevance. Profiles are generated through the configuration of techniques for resource sampling from datasets, topic extraction from reference datasets and their ranking based on graphical models. To enable a good trade-off between scalability and accuracy of generated profiles, appropriate parameters are determined experimentally. Our evaluation considers topic profiles for all accessible datasets from the Linked Open Data cloud. The results show that our approach generates accurate profiles even with comparably small sample sizes (10%) and outperforms established topic modelling approaches.
The term Linked Data refers to a set of best practices for publishing and connecting structured data on the Web. These best practices have been adopted by an increasing number of data providers over the last three years, leading to the creation of a global data space containing billions of assertions-the Web of Data. In this article we present the concept and technical principles of Linked Data, and situate these within the broader context of related technological developments. We describe progress to date in publishing Linked Data on the Web, review applications that have been developed to exploit the Web of Data, and map out a research agenda for the Linked Data community as it moves forward.
Data profiling comprises a broad range of methods to efficiently analyze a given data set. In a typical scenario, which mirrors the capabilities of commercial data profiling tools, tables of a relational database are scanned to derive metadata, such as data types and value patterns, completeness and uniqueness of columns, keys and foreign keys, and occasionally functional dependencies and association rules. Individual research projects have proposed several additional profiling tasks, such as the discovery of inclusion dependencies or conditional functional dependencies.
Data profiling deserves a fresh look for two reasons: First, the area itself is neither established nor defined in any principled way, despite significant research activity on individual parts in the past. Second, more and more data beyond the traditional relational databases are being created and beg to be profiled. The article proposes new research directions and challenges, including interactive and incremental profiling and profiling heterogeneous and non-relational data.
Large amounts of graph-structured data are emerging from various avenues, ranging from natural and life sciences to social and semantic web communities. We address the problem of discovering subgraphs of entities that reflect latent topics in graph-structured data. These topics are structured meta-information providing further insights into the data. The presented approach effectively detects such topics by exploiting only the structure of the underlying graph, thus avoiding the dependency on textual labels, which are a scarce asset in prevalent graph datasets. The viability of our approach is demonstrated in experiments on real-world datasets.
The DBpedia community project extracts structured, multilingual knowledge from Wikipedia and makes it freely available on the Web using Semantic Web and Linked Data technologies. The project extracts knowledge from 111 different language editions of Wikipedia. The largest DBpedia knowledge base which is extracted from the English edition of Wikipedia consists of over 400 million facts that describe 3.7 million things. The DBpedia knowledge bases that are extracted from the other 110 Wikipedia editions together consist of 1.46 billion facts and describe 10 million additional things. The DBpedia project maps Wikipedia infoboxes from 27 different language editions to a single shared ontology consisting of 320 classes and 1,650 properties. The mappings are created via a world-wide crowd-sourcing effort and enable knowledge from the different Wikipedia editions to be combined. The project publishes releases of all DBpedia knowledge bases for download and provides SPARQL query access to 14 out of the 111 language editions via a global network of local DBpedia chapters. In addition to the regular releases, the project maintains a live knowledge base which is updated whenever a page in Wikipedia changes. DBpedia sets 27 million RDF links pointing into over 30 external data sources and thus enables data from these sources to be used together with DBpedia data. Several hundred data sets on the Web publish RDF links pointing to DBpedia themselves and make DBpedia one of the central interlinking hubs in the Linked Open Data (LOD) cloud. In this system report, we give an overview of the DBpedia community project, including its architecture, technical implementation, maintenance, internationalisation, usage statistics and applications.
To integrate Linked Open Data, which originates from various and heterogeneous sources, the use of well-defined ontologies is essential. However, oftentimes the utilization of these ontologies by data publishers differs from the intended application envisioned by ontology engineers. This may lead to unspecified properties being used ad-hoc as predicates in RDF triples or it may result in infrequent usage of specified properties. These mismatches impede the goals and propagation of the Web of Data as data consumers face difficulties when trying to discover and integrate domain-specific information. In this work, we identify and classify common misusage patterns by employing frequency analysis and rule mining. Based on this analysis, we introduce an algorithm to propose suggestions for a data-driven ontology re-engineering workflow, which we evaluate on two large-scale RDF datasets.
The term "Linked Data" refers to a set of best practices for publishing and connecting structured data on the Web. These best practices have been adopted by an increasing number of data providers over the last three years, leading to the creation of a global data space containing billions of assertions-the Web of Data. In this article, the authors present the concept and technical principles of Linked Data, and situate these within the broader context of related technological developments. They describe progress to date in publishing Linked Data on the Web, review applications that have been developed to exploit the Web of Data, and map out a research agenda for the Linked Data community as it moves forward.
The World-Wide Web consists of a huge number of unstruc- tured hypertext documents, but it also contains structured data in the form of HTML tables. Many of these tables contain both relational-style data and a small "schema" of labeled and typed columns, making each such table a small structured database. The WebTables project is an effort to extract and make use of the huge number of these structured tables on the Web. A clean collection of relational-style ta- bles could be useful for improving web search, schema de- sign, and many other applications. This paper describes the first stage of the WebTables project. First, we give an in-depth study of the Web's HTML table corpus. For example, we extracted 14.1 billion HTML ta- bles from a several-billion-page portion of Google's general- purpose web crawl, and estimate that 154 million of these tables contain high-quality relational-style data. We also de- scribe the crawl's distribution of table sizes and data types. Second, we describe a system for performing relation recov- ery. The Web mixes relational and non-relational tables indiscriminately (often on the same page), so there is no simple way to distinguish the 1.1% of good relations from the remainder, nor to recover column label and type infor- mation. Our mix of hand-written detectors and statistical classifiers takes a raw Web crawl as input, and generates a collection of databases that is five orders of magnitude larger than any other collection we are aware of. Relation recovery achieves precision and recall that are comparable to other domain-independent information extraction systems.
Recently, the opportunity of extracting structured data from the Web has been identified by a number of research projects. One such example is that millions of relational-style HTML tables can be extracted from the Web. Traditional data integration approaches do not scale over such corpora with hundreds of small tables in one domain. To solve this problem, previous work has proposed pay-as-you-go data integration systems to provide, with little up-front cost, base services over loosely-integrated informa- tion. One key component of such systems, which has received little attention to date, is the need for a framework to gauge and improve the quality of the integration. We propose a frame- work based on functional dependencies(FDs). Unlike in tradi- tional database design, where FDs are specified as statements of truth about all possible instances of the database; in web envi- ronment, FDs are not specified over the data tables. Instead, we generate FDs by counting-based algorithms over many data sources, and extend the FDs with probabilities to capture the in- herent uncertainties in them. Given these probabilistic FDs, we show how to solve two problems to improve data and schema qual- ity in a pay-as-you-go system: (1) pinpointing dirty data sources and (2) normalizing large mediated schemas. We describe these techniques and evaluate them over real-world data sets extracted from the Web.
Over the last three years, an increasing number of data providers have started to publish structured data according to the Linked Data principles on the Web. The resulting Web of Data currently consists of over 28 billion RDF triples. As the Web of Data grows, there is an increasing need for link discovery tools which scale to very large datasets. In record linkage, many partitioning methods have been proposed which substantially reduce the number of required entity comparisons. Unfortunately, most of these methods either lead to a decrease in recall or only work on metric spaces. We propose a novel blocking method called Multi-Block which uses a multidimensional index in which similar objects are located near each other. In each dimension the entities are indexed by a different property increasing the efficiency of the index significantly. In addition, it guarantees that no false dismissals can occur. Our approach works on complex link specifications which aggregate several different similarity measures. MultiBlock has been implemented as part of the Silk Link Discovery Framework. The evaluation shows a speedup factor of several 100 for large datasets compared to the full evaluation without losing recall.
A statistical profile summarizes the instances of a database. It describes aspects such as the number of tuples, the number of values, the distribution of values, the correlation between value sets, and the distribution of tuples among secondary storage units. Estimation of database profiles is critical in the problems of query optimization, physical database design, and database performance prediction. This paper describes a model of a database of profile, relates this model to estimating the cost of database operations, and surveys methods of estimating profiles. The operators and objects in the model include build profile, estimate profile, and update profile. The estimate operator is classified by the relational algebra operator (select, project, join), the property to be estimated (cardinality, distribution of values, and other parameters), and the underlying method (parametric, nonparametric, and ad-hoc). The accuracy, overhead, and assumptions of methods are discussed in detail. Relevant research in both the database and the statistics disciplines is incorporated in the detailed discussion.
One of the main challenges that the Semantic Web faces is the integration of
a growing number of independently designed ontologies. In this work, we present
PARIS, an approach for the automatic alignment of ontologies. PARIS aligns not
only instances, but also relations and classes. Alignments at the instance
level cross-fertilize with alignments at the schema level. Thereby, our system
provides a truly holistic solution to the problem of ontology alignment. The
heart of the approach is probabilistic, i.e., we measure degrees of matchings
based on probability estimates. This allows PARIS to run without any parameter
tuning. We demonstrate the efficiency of the algorithm and its precision
through extensive experiments. In particular, we obtain a precision of around
90% in experiments with some of the world's largest ontologies.
The Xeb graph is a giaut social network whose properties have been measured and modeled extensively in recent years. Most such studies concentrate on the graph structure alone, arid do not consider textual properties of the riodes. Consequently, Web communities have been characterized purely in terms of graph structure and not on page content. We propose that a topic taxonomy such as Yahoo! or the Open Directory provides a useful flamework for understanding the structure of content-based clusters arid communities. In particular, using a topic taxonomy and an automatic classifier, we can measure the background distribution of broad topics on the Web, and analyze the capability of recent random walk algorithms to draw samples which follow such distributions. In addition, we can measm'e the probability that a page about one broad topic will link to another broad topic. Extending this experiment, we can measure how quickly topic context is lost while walking randomly on the Web graph. Estimates of this topic mixing distance may explain why a global PageRank is still meaningful in the context of broad queries. In general, our measurements may prove valuable in the design of community-specific crawlers and link-based ranking systems.
Cross-domain knowledge bases such as DBpedia, YAGO, or the Google Knowledge Graph have gained increasing attention over the last years and are starting to be deployed within various use cases. However, the content of such knowledge bases is far from being complete, far from always being correct, and suffers from deprecation (i.e. population numbers become outdated after some time). Hence, there are efforts to leverage various types of Web data to complement, update and extend such knowledge bases. A source of Web data that potentially provides a very wide coverage are millions of relational HTML tables that are found on the Web. The existing work on using data from Web tables to augment cross-domain knowledge bases reports only aggregated performance numbers. The actual content of the Web tables and the topical areas of the knowledge bases that can be complemented using the tables remain unclear. In this paper, we match a large, publicly available Web table corpus to the DBpedia knowledge base. Based on the matching results, we profile the potential of Web tables for augmenting different parts of cross-domain knowledge bases and report detailed statistics about classes, properties, and instances for which missing values can be filled using Web table data as evidence. In order to estimate the potential quality of the new values, we empirically examine the Local Closed World Assumption and use it to determine the maximal number of correct facts that an ideal data fusion strategy could generate. Using this as ground truth, we compare three data fusion strategies and conclude that knowledge-based trust outperforms PageRank- and voting-based fusion.
Schema.org is a way to add machine-understandable information to web pages that is processed by the major search engines to improve search performance. The definition of schema.org is provided as a set of web pages plus a partial mapping into RDF triples with unusual properties, and is incomplete in a number of places. This analysis of and formal semantics for schema.org provides a complete basis for a plausible version of what schema.org should be.
In order to support web applications to understand the content of HTML pages an increasing number of websites have started to annotate structured data within their pages using markup formats such as Microdata, RDFa, Microformats. The annotations are used by Google, Yahoo!, Yandex, Bing and Facebook to enrich search results and to display entity descriptions within their applications. In this paper, we present a series of publicly accessible Microdata, RDFa, Microformats datasets that we have extracted from three large web corpora dating from 2010, 2012 and 2013. Altogether, the datasets consist of almost 30 billion RDF quads. The most recent of the datasets contains amongst other data over 211 million product descriptions, 54 million reviews and 125 million postal addresses originating from thousands of websites. The availability of the datasets lays the foundation for further research on integrating and cleansing the data as well as for exploring its utility within different application contexts. As the dataset series covers four years, it can also be used to analyze the evolution of the adoption of the markup formats.
Being promoted by major search engines such as Google, Yahoo!, Bing, and Yandex, Microdata embedded in web pages, especially using schema.org, has become one of the most important markup languages for the Web. However, deployed Microdata is most often not free from errors, which limits its practical use. In this paper, we use the WebDataCommons corpus of Microdata extracted from more than 250 million web pages for a quantitative analysis of common mistakes in Microdata provision. Since it is unrealistic that data providers will provide clean and correct data, we discuss a set of heuristics that can be applied on the data consumer side to fix many of those mistakes in a post-processing step. We apply those heuristics to provide an improved knowledge base constructed from the raw Microdata extraction.
Semantically annotated data, using markup languages like RDFa and Microdata, has become more and more publicly available in the Web, especially in the area of e-commerce. Thus, a large amount of structured product descriptions are freely available and can be used for various applications, such as product search or recommendation. However, little efforts have been made to analyze the categories of the available product descriptions. Although some products have an explicit category assigned, the categorization schemes vary a lot, as the products originate from thousands of different sites. This heterogeneity makes the use of supervised methods, which have been proposed by most previous works, hard to apply. Therefore, in this paper, we explain how distantly supervised approaches can be used to exploit the heterogeneous category information in order to map the products to set of target categories from an existing product catalogue. Our results show that, even though this task is by far not trivial, we can reach almost accuracy for classifying products into 37 categories.
Promoted by major search engines, schema.org has become a widely adopted standard for marking up structured data in HTML web pages. In this paper, we use a series of large-scale Web crawls to analyze the evolution and adoption of schema.org over time. The availability of data from different points in time for both the schema and the websites deploying data allows for a new kind of empirical analysis of standards adoption, which has not been possible before. To conduct our analysis, we compare different versions of the schema.org vocabulary to the data that was deployed on hundreds of thousands of Web pages at different points in time. We measure both top-down adoption (i.e., the extent to which changes in the schema are adopted by data providers) as well as bottom-up evolution (i.e., the extent to which the actually deployed data drives changes in the schema). Our empirical analysis shows that both processes can be observed.
Millions of HTML tables containing structured data can be found on the Web. With their wide coverage, these tables are potentially very useful for filling missing values and extending cross-domain knowledge bases such as DBpedia, YAGO, or the Google Knowledge Graph. As a prerequisite for being able to use table data for knowledge base extension, the HTML tables need to be matched with the knowledge base, meaning that correspondences between table rows/columns and entities/schema elements of the knowledge base need to be found. This paper presents the T2D gold standard for measuring and comparing the performance of HTML table to knowledge base matching systems. T2D consists of 8 700 schema-level and 26 100 entity-level correspondences between the WebDataCommons Web Tables Corpus and the DBpedia knowledge base. In contrast related work on HTML table to knowledge base matching, the Web Tables Corpus (147 million tables), the knowledge base, as well as the gold standard are publicly available. The gold standard is used afterward to evaluate the performance of T2K Match, an iterative matching method which combines schema and instance matching. T2K Match is designed for the use case of matching large quantities of mostly small and narrow HTML tables against large cross-domain knowledge bases. The evaluation using the T2D gold standard shows that T2K Match discovers table-to-class correspondences with a precision of 94%, row-to-entity correspondences with a precision of 90%, and column-to-property correspondences with a precision of 77%.
The central idea of Linked Data is that data publishers support applications in discovering and integrating data by complying to a set of best practices in the areas of linking, vocabulary usage, and metadata provision. In 2011, the State of the LOD Cloud report analyzed the adoption of these best practices by linked datasets within different topical domains. The report was based on information that was provided by the dataset publishers themselves via the datahub.io Linked Data catalog. In this paper, we revisit and update the findings of the 2011 State of the LOD Cloud report based on a crawl of the Web of Linked Data conducted in April 2014. We analyze how the adoption of the different best practices has changed and present an overview of the linkage relationships between datasets in the form of an updated LOD cloud diagram, this time not based on information from dataset providers, but on data that can actually be retrieved by a Linked Data crawler. Among others, we find that the number of linked datasets has approximately doubled between 2011 and 2014, that there is increased agreement on common vocabularies for describing certain types of entities, and that provenance and license metadata is still rarely provided by the data sources.
The web is a rich resource of structured data. There has been an increasing interest in using web structured data for many applications such as data integration, web search and question answering. In this paper, we present DEXTER, a system to find product sites on the web, and detect and extract product specifications from them. Since product specifications exist in multiple product sites, our focused crawler relies on search queries and backlinks to discover product sites. To perform the detection, and handle the high diversity of specifications in terms of content, size and format, our system uses supervised learning to classify HTML fragments (e.g., tables and lists) present in web pages as specifications or not. To perform large-scale extraction of the attribute-value pairs from the HTML fragments identified by the specification detector, DEXTER adopts two lightweight strategies: a domain-independent and unsupervised wrapper method, which relies on the observation that these HTML fragments have very similar structure; and a combination of this strategy with a previous approach, which infers extraction patterns by annotations generated by automatic but noisy annotators. The results show that our crawler strategy to locate product specification pages is effective: (1) it discovered 1:46M product specification pages from 3; 005 sites and 9 different categories; (2) the specification detector obtains high values of F-measure (close to 0:9) over a heterogeneous set of product specifications; and (3) our efficient wrapper methods for attribute-value extraction get very high values of precision (0.92) and recall (0.95) and obtain better results than a state-of-the-art, supervised rule-based wrapper.
The web is a rich resource of structured data. There has been an increasing interest in using web structured data for many applications such as data integration, web search and question answering. In this paper, we present Dexter, a system to find product sites on the web, and detect and extract product specifications from them. Since product specifications exist in multiple product sites, our focused crawler relies on search queries and backlinks to discover product sites. To perform the detection, and handle the high diversity of specifications in terms of content, size and format, our system uses supervised learning to classify HTML fragments (e.g., tables and lists) present in web pages as specifications or not. To perform large-scale extraction of the attribute-value pairs from the HTML fragments identified by the specification detector, Dexter adopts two lightweight strategies: a domain-independent and unsupervised wrapper method, which relies on the observation that these HTML fragments have very similar structure; and a combination of this strategy with a previous approach, which infers extraction patterns by annotations generated by automatic but noisy annotators. The results show that our crawler strategy to locate product specification pages is effective: (1) it discovered 1:46AM product specification pages from 3; 005 sites and 9 different categories; (2) the specification detector obtains high values of F-measure (close to 0:9) over a heterogeneous set of product specifications; and (3) our efficient wrapper methods for attribute-value extraction get very high values of precision (0.92) and recall (0.95) and obtain better results than a state-of-the-art, supervised rule-based wrapper.
Knowledge about the general graph structure of the World Wide Web is important for understanding the social mechanisms that govern its growth, for designing ranking methods, for devising better crawling algorithms, and for creating accurate models of its structure. In this paper, we analyze a large web graph. The graph was extracted from a large publicly accessible web crawl that was gathered by the Common Crawl Foundation in 2012. The graph covers over 3:5 billion web pages and 128:7 billion links. We analyse and compare, among other features, degree distributions, connectivity, average distances, and the structure of weakly/strongly connected components. We conduct our analysis on three different levels of aggregation: page, host, and pay-level domain (PLD) (one “dot level” above public suffixes). Our analysis shows that, as evidenced by previous research , some of the features previously observed by Broder et al. are very dependent on artifacts of the crawling process, whereas other appear to be more structural. We confirm the existence of a giant strongly connected component; we however find, as observed by other researchers, very different proportions of nodes that can reach or that can be reached from the giant component, suggesting that the “bow-tie structure” as described in is strongly dependent on the crawling process, and to the best of our current knowledge is not a structural property of the web. More importantly, statistical testing and visual inspection of sizerank plots show that the distributions of indegree, outdegree and sizes of strongly connected components of the page and host graph are not power laws, contrarily to what was previously reported for much smaller crawls, although they might be heavy tailed. If we aggregate at pay-level domain, however, a power law emerges. We also provide for the first time accurate measurement of distancebased features, using recently introduced algorithms that scale to the size of our crawl.
The development of relational database management systems served to focus the data management community for decades, with spectacular results. In recent years, however, the rapidly-expanding demands of "data everywhere" have led to a field comprised of interesting and productive efforts, but without a central focus or coordinated agenda. The most acute information management challenges today stem from organizations (e.g., enterprises, government agencies, libraries, "smart" homes) relying on a large number of diverse, interrelated data sources, but having no way to manage their dataspaces in a convenient, integrated, or principled fashion. This paper proposes dataspaces and their support systems as a new agenda for data management. This agenda encompasses much of the work going on in data management today, while posing additional research objectives.
A Search Join is a join operation which extends a user-provided table with additional attributes based on a large corpus of heterogeneous data originating from the Web or corporate intranets. Search Joins are useful within a wide range of application scenarios: Imagine you are an analyst having a local table describing companies and you want to extend this table with attributes containing the headquarters, turnover, and revenue of each company. Or imagine you are a film enthusiast and want to extend a table describing films with attributes like director, genre, and release date of each film. This article presents the Mannheim Search Join Engine which automatically performs such table extension operations based on a large corpus of Web data. Given a local table, the Mannheim Search Join Engine searches the corpus for additional data describing the entities contained in the input table. The discovered data are joined with the local table and are consolidated using schema matching and data fusion techniques. As result, the user is presented with an extended table and given the opportunity to examine the provenance of the added data. We evaluate the Mannheim Search Join Engine using heterogeneous data originating from over one million different websites. The data corpus consists of HTML tables, as well as Linked Data and Microdata annotations which are converted into tabular form. Our experiments show that the Mannheim Search Join Engine achieves a coverage close to 100% and a precision of around 90% for the tasks of extending tables describing cities, companies, countries, drugs, books, films, and songs.
We propose a method for consolidating entities in RDF data on the Web. Our approach is based on a statistical analysis of the use of predicates and their associated values to identify "quasi"-key proper-ties. Compared to a purely symbolic based approach, we obtain promising results, retrieving more identical entities with a high precision. We also argue that our technique scales well—possibly to the size of the current Web of Data—as opposed to more expensive existing approaches.
We provide an analysis of the adoption of metadata stan- dards on the Web based a large crawl of the Web. In particular, we look at what forms of syntax and vocabularies publishers are using to mark up data inside HTML pages. We also describe the process that we have followed and the dificulties involved in web data extraction.
More and more websites embed structured data describing for instance products, reviews, blog posts, people, organizations, events, and cooking recipes into their HTML pages using markup standards such as Microformats, Microdata and RDFa. This development has accelerated in the last two years as major Web companies, such as Google, Facebook, Yahoo!, and Microsoft, have started to use the embedded data within their applications. In this paper, we analyze the adoption of RDFa, Microdata, and Microformats across the Web. Our study is based on a large public Web crawl dating from early 2012 and consisting of 3 billion HTML pages which originate from over 40 million websites. The analysis reveals the deployment of the different markup standards, the main topical areas of the published data as well as the different vocabularies that are used within each topical area to represent data. What distinguishes our work from earlier studies, published by the large Web companies, is that the analyzed crawl as well as the extracted data are publicly available. This allows our findings to be verified and to be used as starting points for further domain-specific investigations as well as for focused information extraction endeavors.
Large numbers of websites have started to markup their content using standards such as Microdata, Microformats, and RDFa. The marked-up content elements comprise descriptions of people, organizations, places, events, products, ratings, and reviews. This development has accelerated in last years as major search engines such as Google, Bing and Yahoo! use the markup to improve their search results. Embedding semantic markup facilitates identifying content elements on webpages. However, the markup is mostly not as fine-grained as desirable for applications that aim to integrate data from large numbers of websites. This paper discusses the challenges that arise in the task of integrating descriptions of electronic products from several thousand e-shops that offer Microdata markup. We present a solution for each step of the data integration process including Microdata extraction, product classification, product feature extraction, identity resolution, and data fusion. We evaluate our processing pipeline using 1.9 million product offers from 9240 e-shops which we extracted from the Common Crawl 2012, a large public Web corpus.
A central problem in data integration and data cleansing is to find entities
in different data sources that describe the same real-world object. Many
existing methods for identifying such entities rely on explicit linkage rules
which specify the conditions that entities must fulfill in order to be
considered to describe the same real-world object. In this paper, we present
the GenLink algorithm for learning expressive linkage rules from a set of
existing reference links using genetic programming. The algorithm is capable of
generating linkage rules which select discriminative properties for comparison,
apply chains of data transformations to normalize property values, choose
appropriate distance measures and thresholds and combine the results of
multiple comparisons using non-linear aggregation functions. Our experiments
show that the GenLink algorithm outperforms the state-of-the-art genetic
programming approach to learning linkage rules recently presented by Carvalho
et. al. and is capable of learning linkage rules which achieve a similar
accuracy as human written rules for the same problem.
There has been a recent, tangible growth in RDF published on the Web in accordance with the Linked Data principles and best practices, the result of which has been dubbed the “Web of Data”. Linked Data guidelines are designed to facilitate ad hoc re-use and integration of conformant structured data–across the Web–by consumer applications; however, thus far, systems have yet to emerge that convincingly demonstrate the potential applications for consuming currently available Linked Data. Herein, we compile a list of fourteen concrete guidelines as given in the “How to Publish Linked Data on the Web” tutorial. Thereafter, we evaluate conformance of current RDF data providers with respect to these guidelines. Our evaluation is based on quantitative empirical analyses of a crawl of ∼4 million RDF/XML documents constituting over 1 billion quadruples, where we also look at the stability of hosted documents for a corpus consisting of nine monthly snapshots from a sample of 151 thousand documents. Backed by our empirical survey, we provide insights into the current level of conformance with respect to various Linked Data guidelines, enumerating lists of the most (non-)conformant data providers. We show that certain guidelines are broadly adhered to (esp. use HTTP URIs, keep URIs stable), whilst others are commonly overlooked (esp. provide licencing and human-readable meta-data). We also compare PageRank scores for the data-providers and their conformance to Linked Data guidelines, showing that both factors negatively correlate for guidelines restricting the use of RDF features, while positively correlating for guidelines encouraging external linkage and vocabulary re-use. Finally, we present a summary of conformance for the different guidelines, and present the top-ranked data providers in terms of a combined PageRank and Linked Data conformance score.
Web search services are among the most heavily used applications on the World Wide Web. Perhaps because search is used in such a huge variety of tasks and contexts, the user interface must strike a careful balance to meet all user needs. We describe a study that used eye tracking methodologies to explore the effects of changes in the presentation of search results. We found that adding information to the contextual snippet significantly improved performance for informational tasks but degraded performance for navigational tasks. We discuss possible reasons for this difference and the design implications for better presentation of search results.
The focus of web search is moving away from returning rel- evant documents towards returning structured data as results to user queries. A vital part in the architecture of search engines are link-based ranking algorithms, which however are targeted towards hypertext docu- ments. Existing ranking algorithms for structured data, on the other hand, require manual input of a domain expert and are thus not applicable in cases where data integrated from a large number of sources exhibits enor- mous variance in vocabularies used. In such environments, the authority of data sources is an important signal that the ranking algorithm has to take into account. This paper presents algorithms for prioritising data returned by queries over web datasets expressed in RDF. We introduce the notion of naming authority which provides a correspondence between identifiers and the sources which can speak authoritatively for these identifiers. Our algorithm uses the original PageRank method to assign authority values to data sources based on a naming authority graph, and then propagates the authority values to identifiers referenced in the sources. We conduct performance and quality evaluations of the method on a large web dataset. Our method is schema-independent, requires no manual input, and has ap- plications in search, query processing, reasoning, and user interfaces over integrated datasets.
Over a decade after RDF has been published as a W3C recommendation, publishing open and machine-readable content on the Web has recently received a lot more attention, including from corporate and governmental bodies; notably thanks to the Linked Open Data community, there now exists a rich vein of heterogeneous RDF data published on the Web (the so-called "Web of Data") accessible to all. However, RDF publishers are prone to making errors which compromise the effectiveness of applications leveraging the resulting data. In this paper, we discuss common errors in RDF publishing, their consequences for applications, along with possible publisher-oriented approaches to improve the quality of structured, machine-readable and open data on the Web.
In this paper we report the results of an experimental study on the validation problem of existing HTML pages in the World Wide Web. We have found that only about 5% of webpages are "valid" according to the HTML standard. An "invalid" webpage may be rendered differently by different browsers; it may not be machine- processable; it might not be translated into other Web document formats correctly using some of the translation engines available on the Web; etc. Through sampling and analyzing the results, we have identified the most common problems in existing webpages that made them invalid. We hope our discovery can encourage standard bodies, browser developers, authoring tool developers, and webpage designers to work together so that the number of valid webpages continues to grow and the Web can indeed reach its full potential.
Web pages often contain clutter (such as pop-up ads, unnecessary images and extraneous links) around the body of an article that distracts a user from actual content. Extraction of "useful and relevant" content from web pages has many applications, including cell phone and PDA browsing, speech rendering for the visually impaired, and text summarization. Most approaches to removing clutter or making content more readable involve changing font size or removing HTML and data components such as images, which takes away from a webpage's inherent look and feel. Unlike "Content Reformatting", which aims to reproduce the entire webpage in a more convenient form, our solution directly addresses "Content Extraction". We have developed a framework that employs easily extensible set of techniques that incorporate advantages of previous work on content extraction. Our key insight is to work with the DOM trees, rather than with raw HTML markup. We have implemented our approach in a publicly available Web proxy to extract content from HTML web pages.
What the adoption of schema.org tells about linked open data
- Heiko Paulheim
Analyzing Schema.org, Proceedings of the 13th International Semantic Web Conference-Part I
- F Peter
- Patel-Schneider
Microformats and RDFa deployment across the Web
- Peter Mika