From d4bbed4e75b4cfa91967237ddb13b5b6321b3328 Mon Sep 17 00:00:00 2001 From: Artem V L Date: Mon, 9 Apr 2018 16:44:03 +0200 Subject: [PATCH] Evaluation description refined --- README.md | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/README.md b/README.md index 9a606be..beb0aca 100644 --- a/README.md +++ b/README.md @@ -25,13 +25,14 @@ For the semi supervised mode, similarity between the RDF subjects is evaluated w > Currently StaTIX infers semantic types based only on the statistical analysis of the input dataset. The accuracy can be improved further with additional semantic analyses leveraging both logical inference and embedding techniques to better grasp the differences and relationships between various instances. - The output results are clusters in the [.cnl format](https://github.com/eXascaleInfolab/PyCABeM/blob/master/formats/format.cnl) (space separated list of members). Each cluster correspond to the type and has members represented by the subject ids. Subject ids are generated sequentially starting from `0` for all unique subjects in the input dataset. -Evaluation of the StaTIX type inference (see details in [TInfES](https://github.com/eXascaleInfolab/TInfES)): -Accuracy (the higher, the better): +### Evaluation of the StaTIX type inference (see details in [TInfES](https://github.com/eXascaleInfolab/TInfES)) +StaTIX is executed as `./run.sh -f -m -r 1 -o ` and compared with [SDA](https://link.springer.com/chapter/10.1007/978-3-319-25264-3_36) and [SDType](http://www.heikopaulheim.com/docs/iswc2013.pdf). + +**Accuracy** (the higher, the better): ![F1h](images/F1h_Algs.png) -Execution time (the lower, the better): +**Execution time** (the lower, the better): ![Execution Time](images/ETime_Algs.png) ## Requirements