# DBpedia model artifact
Contains models of DBpedia Spotlight for different languages.

DBpedia model artifact version 2022.03.01

 + ## Update, September 2020
This update generates and upload to dbpedia databus the artifacts of dbpedia-models and wikistatsextractor for all langauges.
Contains some fixes from [Klaus82](https://github.com/dbpedia-spotlight/model-quickstarter/pull/20)
The update is available in [Julio-Noe's Github repo](https://github.com/Julio-Noe/model-quickstarter) and will be merged soon into the spotlight repo.

## wikistatextractor (spotlight-wikistats)

The [wikistatsextractor](https://github.com/dbpedia-spotlight/wikistatsextractor/) extracts statistics statistics from Wikipedia Dump files. It extracts the same 4 files initially produced by pignlproc for dbpedia spotlight.

File  | Long name  |  Line Format
------------- | ------------- | -------------
uriCount  | Articles counts |  ```<title>\t<uri>\t<count>```
pairCount  | Pair-wise occurrence counts |  ```<surface form>\t<uri>\t<count>```
sfAndTotalCount  | Surface forms counts |  ```<surface form>\t<count as SF>\t<count as token>```
tokenCount  | Context vectors |  ```<uri (wikipedia style)>\t{(context_token1,count1),(context_token2, count2),... }```

These files are now available to download in the Databus spotlight-wikistats repository. A short description of each file is as follows (examples could be found in the [DBpedia wiki](https://github.com/dbpedia-spotlight/dbpedia-spotlight/wiki/Raw-data)):

- uriCount: Contains the number of mentions for each article
- pairCount: Defines the number of times a surface form (plain text) was used to reference a resource (DBpedia URI)
- sfAndTotalCount: Establish how many times a surface form was used as an anchor and how many times it occurred in text (regardless of wheter that was part of an anchor). 

### model-quickstarter (spotlight-model)

The dbpedia-models was generated with the [DBpedia model-quickstarter](https://github.com/dbpedia-spotlight/model-quickstarter) tool. 

## Citation

If you use the current (statistical version) of DBpedia Spotlight or the data/models created using this repository, please cite the following paper.

```bibtex
@inproceedings{isem2013daiber,
  title = {Improving Efficiency and Accuracy in Multilingual Entity Extraction},
  author = {Joachim Daiber and Max Jakob and Chris Hokamp and Pablo N. Mendes},
  year = {2013},
  booktitle = {Proceedings of the 9th International Conference on Semantic Systems (I-Semantics)}
}
```