Stuck on an issue?

Lightrun Answers was designed to reduce the constant googling that comes with debugging 3rd party libraries. It collects links to all the places you might be looking at while hunting down a tough bug.

And, if you’re still stuck at the end, we’re happy to hop on a call to see how we can help out.

Reproduce Kinship, FB15k-237 results

See original GitHub issue

Hi @rajarshd, @shehzaadzd,

Using the config for kinship and fb15k-237, we only get the following results, which are quite different from the ones in the paper.

FB15k-237
[ Hits@1:  0.1511 ]
[ Hits@3:  0.2183 ]
[ Hits@5:  0.2620 ]
[ Hits@10:  0.3275 ]
[ Hits@20:  0.3898 ]
[ auc:  0.2048 ]

Kinship
[ Hits@1:  0.2607 ]
[ Hits@3:  0.4981 ]
[ Hits@5:  0.5819 ]
[ Hits@10:  0.6853 ]
[ Hits@20:  0.7551 ]
[ auc:  0.4048 ]

Related to https://github.com/shehzaadzd/MINERVA/issues/10, is there any fixes to reproduce the results?

Issue Analytics

State:
Created 5 years ago
Comments:5

Top GitHub Comments

2reactions

rajarshdcommented, May 9, 2018

Hi @posenhuang, thanks for using our code. Little occupied with the upcoming deadlines. We will try to get back to you by tonight!

0reactions

renhualicommented, Dec 10, 2020

Is this problem solved? I have the same problem，using the config for kinship and fb15k-237, I only get the following results, FB15k-237: Score for iteration 100 Hits@1: 0.0358 Hits@3: 0.0794 Hits@5: 0.1077 Hits@10: 0.1526 Hits@20: 0.1980 auc: 0.0710

Kinship: Score for iteration 1900 Hits@1: 0.4888 Hits@3: 0.7753 Hits@5: 0.8586 Hits@10: 0.9120 Hits@20: 0.9419 auc: 0.6453 Why is the result different from that in the paper?

Top Results From Across the Web

CP Tensor Factorization for Knowledge Graph Completion

KinShip. Table 3. Link prediction results on FB15K237 and WN18RR. FB15K237. WN18RR. MRR Hit@10 Hit@3 Hit@1 MRR Hit@10 Hit@3 Hit@1.

Alliance theory - kinship - Britannica

Lévi-Strauss suggested that, because women's fertility is necessary to the reproduction of the group, women are the “supreme gift.” With no fair return...

KRACL: Contrastive Learning with Graph Context Modeling for ...

Our experiments demonstrate that KRACL achieves superior results across various ... FB15k-237 fb15k237 , WN18RR conve , NELL-995 nell995 , Kinship kinship ......

Emergence of kinship structures and descent systems - PubMed

Anthropologists have revealed several kinship structures. Here, we propose an agent-based model of indigenous societies to reveal the evolution ...

(PDF) Contextual Parameter Generation for Knowledge Graph Link ...

Experimental results on a human-annotated dataset and two distantly ... degree of 0.0 Kinship UMLS FB15k237 WN18RR NELL-995 the graph nodes in the...