Michael
Färber, Frederic Bartscherer, Carsten Menne, and Achim Rettinger
Supplementary
Material for the SWJ Submission
Linked Data Quality of DBpedia,
Freebase, OpenCyc, Wikidata, and
YAGO
Abstract:
In recent years, several noteworthy large, cross-domain and
openly available knowledge graphs (KGs) have been created. These
include DBpedia, Freebase, OpenCyc, Wikidata, and YAGO. Although
extensively in use, these KGs have not been subject to an in-depth
comparison so far. In this survey, we provide data quality criteria
according to which KGs can be analyzed and analyze and compare the
above mentioned KGs. Furthermore, we propose a framework for finding
the most suitable KG for a given setting.The latest submission can be found at
http://www.semantic-web-journal.net/content/linked-data-quality-dbpedia-freebase-opencyc-wikidata-and-yago-0
The final print version of this article will be published soon.
1. KG
Recommendation Framework
The KG
Recommendation Framework as Excel Sheet can be used for selecting
the most suitable KG for a given setting. Given the metric values
regarding the KGs DBpedia, Freebase, OpenCyc, Wikidata, and YAGO (see
our article), the user can weight the metrics according to his/her
needs and get an indication which KGs might fit best.2. Key Statistics
- Data about classes: Contains list of classes KG, ordered by number of instances
- Data about predicates (Pimp): Contains list of predicates of KG (not explicitly defined), ordered by frequency
- Data about relations (Pexp): Contains list of relations of KG (explicitly defined), ordered by frequency
3. Quality Assessment
We provide our data both as tab-separated file (.tsv) and Excel file (.xlsx).- Data for mavail:
Contains downtime and uptime measurements per KG
- Availability of DBpedia:
- Availability of Freebase:
- Availability of OpenCyc:
- Availability of Wikidata:
- Availability of YAGO:
- Data for msemTriple: Contains checked triples (semantic accuracy) about names of people per KG, and for dates and population only the errors (i.e., wrong birth dates in Wikidata and YAGO)
- Data for msynLit: Contains checked literals (syntactic accuracy) for date of Birth, ISBN, population per KG
- Data for mcCol: Contains considered combinations of classes and relations and their filling degrees for measuring column completness per KG
- Data for mconClass: Contains all owl:disjointWith class restriction and their inconsistencies
- Data for mconRelatFunctional: Contains owl:FunctionalProperty relation restriction and how often a relation is used more than once per resource
- Data for mconRelatRange: Contains considered combination of relations and datatypes and how often the defined datatype is not the same on instance level
- Data for mcPop: Contains gold standard (entites) that was checked and the result per KG
- Data for mcSchema: Contains gold standard (classes, relations) that was checked and result per KG
- Data for mDeref: Contains result w.r.t. dereferencing internal URIs by (s,p,o)-position and corresponding HTTP status codes
- Data for mURIs: Containts result w.r.t. validity of external URIs and corresponding HTTP status codes
2. Gold Standard
We use the following gold standard regarding the metrics mcSchema, mcCol, and mcPop.Note that "wn:" stands for the namespace http://wordnet-rdf.princeton.edu/wn31/
The gold standard is also contained in the data for mcPop.
| Domain | Main classes (subclasses in brackets) | Relations (for mcCol) |
|---|---|---|
| Person | Person (wn:100007846-n) (Musician, Athlete, Writer, Politician) |
Person: Date of birth, Place of birth, Gender, Parents (or Mother and Father) |
| Media | Show (wn:100521313-n) (Film, TV series) Literary Composition (wn:106375736-n) (Book, Newspaper, Magazine) Musical Composition (wn:107051211-n) (Album, Song) |
Film: Director, Runtime, Release date TV series: producer Book: Author, ISBN, Number of Pages Album: Artist, Tracklist |
| Organisation | Company (wn:108074934-n) (Bank, Airline) Educational Institution (wn:108293263-n) (University, School) Social Groups (wn:107967506-n) (Sports Club, Political Party) |
Company, University: Location, Founding Year Company: Revenue |
| Geography | Topogaphy (wn:106132185-n) (Lake, River, Mountain) Geographical Area (wn:108693705-n) (Country, Region, City) |
Country: Capital, Area, GDP City: Population, Geolocation |
| Biology | Animal (wn:100015568-n) (Mammal, Bird, Fish, Insect) Plant (wn:100017402-n) (Tree, Grass, Shrub) |
Animal, Plant: Taxonomic Rank |
| Domain | Main class | Sub class | Entity | Short/long |
|---|---|---|---|---|
| Person | Person | Business man | Carlos Slim | s |
| Bill Gates | s | |||
| Karl Albrecht | l | |||
| Michael Otto | l | |||
| Musician | Elvis Presley | s | ||
| Michael Jackson | s | |||
| Peter Maffay | l | |||
| Xavier Naidoo | l | |||
| Athlete | Michael Phelps | s | ||
| Larisa Latynina | s | |||
| Maria Riesch/Höfl-Riesch | l | |||
| Magdalena Neuner | l | |||
| Writer | William Shakespeare | s | ||
| Agatha Christie | s | |||
| Tommy Jaud | l | |||
| Dora Heldt | l | |||
| Politican | Barack Obama | s | ||
| George W. Bush | s | |||
| Stefan Mappus | l | |||
| Günther Oettinger | l | |||
| Media | Show | Film | Avatar | s |
| Titanic | s | |||
| Friendship! | l | |||
| Konfernz der Tiere/Animals United | l | |||
| TV Series | The Walking Dead | s | ||
| Full House | s | |||
| Tatort | l | |||
| Alarm für Cobra 11 | l | |||
| Literary Composition | Book | The Lord of the Rings (Tolkien) | s | |
| The Hobbit (Tolkien) | s | |||
| Hummeldumm (Jaud) | l | |||
| nothing-avail Wort zu Papa (Heldt) | l | |||
| Magazine | Time | s | ||
| Newsweek | s | |||
| TV14 | l | |||
| TV Spielfilm | l | |||
| Musical Composition | Album | M. Jackson - Thriller | s | |
| AC/DC - Back in Black | s | |||
| Unheilig - Grosse Freiheit | l | |||
| Gute Reise - Ich+Ich | l | |||
| Organisation | Company | Bank | JP Morgan Chase | s |
| Bank of America | s | |||
| Landesbank Baden-Württemberg | l | |||
| BayernLB | l | |||
| Airlines | Delta Air Lines | s | ||
| United Airlines | s | |||
| Lufthansa | l | |||
| Air Berlin | l | |||
| Educational Institution | University | California Institute of Technology | s | |
| Harvard University | s | |||
| LMU München / Ludwig Maximilian University of Munich | l | |||
| University of Göttingen | l | |||
| Social Groups | Sports club | Real Madrid | s | |
| FC Barcelona | s | |||
| Hertha BSC | l | |||
| FC Augsburg | l | |||
| Political Party | Republican Party (U.S) | s | ||
| Democratic Party (U.S) | s | |||
| SPD / Social Democratic Party of Germany | l | |||
| CDU / Christian Democratic Union of Germany | l | |||
| Geography | Topography | Lake | Caspian Sea | s |
| Lake Superior (US) | s | |||
| Bodensee / Lake Constance | l | |||
| Müritz | l | |||
| River | Nile | s | ||
| Amazon River | s | |||
| Rhine | l | |||
| Elbe | l | |||
| Mountain | Mount Everest | s | ||
| K2 | s | |||
| Zugspitze | l | |||
| Schneefernerkopf | l | |||
| Country | Russland | s | ||
| Canada | s | |||
| Vatican City | l | |||
| Monaco | l | |||
| City | MexikoStadt | s | ||
| Peking | s | |||
| Stuttgart | l | |||
| Karlsruhe | l | |||
| Biology | Animal | Mammal | Blue whale (Balaenoptera musculus) | s |
| African bush elephant (Loxodonta africana) | s | |||
| Nullarbor dwarf bettong (Bettongia pusilla (EX) | l | |||
| Oriente cave rat (Boromys offella (EX) | l | |||
| Bird | King Pinguin (Aptenodytes patagonicus) | s | ||
| Blue-and-yellow macaw (Ara ararauna) | s | |||
| Bermuda saw-whet owl (Aegolius gradyi (EX) | l | |||
| Mauritius blue pigeon (Alectroenas nitidissimus (EX) | l | |||
| Fish | Great white shark (Carcharodon carcharias) | s | ||
| Red-bellied piranha (Pygocentrus nattereri) | s | |||
| Ornate sleeper-ray (Electrolux addisoni (CR) | l | |||
| Irrawaddy river shark (Glyphis siamensis (CR) | l | |||
| Plant | Tree | Giant Sequoia (Sequoiadendron giganteum) | s | |
| Apple Tree (Malus domestica) | s | |||
| Araucaria nemorosa (CR) | l | |||
| Dacrydium guillauminii (CR) | l | |||
| Grass | Corn/Maize (Zea mays) | s | ||
| Rye (Secale cereale) | s | |||
| Cyperus flavoculmis (CR) | l | |||
| Cyperus microumbellatus (CR) | l |
All provided data regarding the key statistics and quality assessment is also available as one single archive file here.
April 2016