Knowledge graph completion, also known as link prediction, is a task that involves inferring missing information from knowledge graphs. It is widely used in various applications such as product recommendation and question answering. However, existing approaches to knowledge graph completion, which rely solely on the information present in the input knowledge graph, often yield unsatisfactory results. This limitation makes these solutions inadequate for critical domains like healthcare. To address this issue and improve the accuracy of knowledge graph completion, we propose a novel approach that combines the data-driven power of knowledge graph embeddings with domain-specific reasoning derived from experts or entailment regimes like OWL2. By loosely coupling these two techniques, we not only enhance prediction accuracy by incorporating domain knowledge that may not be explicitly included in the input knowledge graph but also allow users to integrate their own knowledge graph embedding and reasoning methods. Our initial results demonstrate significant improvements in mean reciprocal rank (MRR) accuracy compared to vanilla knowledge graph embeddings. We achieve up to 3 times enhancement in MRR accuracy and outperform hybrid solutions that combine knowledge graph embeddings with rule mining and reasoning by up to 3.5 times MRR. In order to validate our proposal, we conducted evaluations using both synthetic and real-world knowledge graphs. The datasets used include DBPedia (a real-world KG accompanied by an ontology) and LUBM (a popular synthetic KG generator). Due to computational limitations, we worked with a subset of DBPedia containing a well-connected subgraph consisting of the 12 most popular relations and related information. We compared our approach against several baseline models including TransE, DistMult, ComplEx, HolE, ConvE, KALE, and plogicNet. For each dataset and model, we utilized consistent hyperparameter settings for both vanilla KGEs (knowledge graph embeddings) and our approach. The hardware used for experimentation consisted of a machine with an AMD Opteron processor running at 32x2.3 GHz, 62GB RAM, and an NVIDIA GPU.
- Knowledge graph completion, also known as link prediction, infers missing information from knowledge graphs
- Existing approaches relying solely on input knowledge graph often yield unsatisfactory results
- Proposed approach combines knowledge graph embeddings with domain-specific reasoning to improve accuracy
- Combination of techniques enhances prediction accuracy by incorporating domain knowledge not explicitly included in the input knowledge graph
- Users can integrate their own knowledge graph embedding and reasoning methods
- Initial results show significant improvements in mean reciprocal rank (MRR) accuracy compared to vanilla knowledge graph embeddings
- Outperforms hybrid solutions that combine knowledge graph embeddings with rule mining and reasoning by up to 3.5 times MRR
- Evaluations conducted using synthetic and real-world knowledge graphs including DBPedia and LUBM datasets
- Subset of DBPedia used due to computational limitations, consisting of a well-connected subgraph with 12 popular relations and related information
- Approach compared against several baseline models including TransE, DistMult, ComplEx, HolE, ConvE, KALE, and plogicNet
- Consistent hyperparameter settings used for both vanilla KGEs (knowledge graph embeddings) and proposed approach
- Experimentation conducted on a machine with an AMD Opteron processor running at 32x2.3 GHz, 62GB RAM, and an NVIDIA GPU.
Knowledge graph completion, or link prediction, is a way to find missing information in knowledge graphs. Some methods that only use the input knowledge graph don't work well. A new method combines knowledge graph embeddings with domain-specific reasoning to make predictions more accurate. This combination of techniques improves accuracy by including extra domain knowledge that isn't in the original graph. Users can also add their own methods for embedding and reasoning. The new method performs better than other methods that combine embeddings with rule mining and reasoning."
Definitions
- Knowledge graph completion: Finding missing information in a knowledge graph.
- Link prediction: Inferring missing information from a knowledge graph.
- Knowledge graphs: Graphs that organize information and show relationships between different pieces of data.
- Embeddings: Representations of data in a lower-dimensional space.
- Domain-specific reasoning: Using specific knowledge about a particular subject to make predictions or decisions.
- Accuracy: How correct or close to the truth something is.
- Baseline models: Standard models used for comparison in experiments.
- Hyperparameter settings: Specific values chosen for certain parameters in an algorithm or model.
- GPU: Graphics Processing Unit, a type of computer hardware used for faster processing.
Knowledge Graph Completion: Combining Knowledge Graph Embeddings with Domain-Specific Reasoning
Knowledge graph completion, also known as link prediction, is a task that involves inferring missing information from knowledge graphs. It is widely used in various applications such as product recommendation and question answering. However, existing approaches to knowledge graph completion often yield unsatisfactory results due to their reliance on the information present in the input knowledge graph only. This limitation makes these solutions inadequate for critical domains like healthcare.
In order to address this issue and improve the accuracy of knowledge graph completion, researchers have proposed a novel approach that combines the data-driven power of knowledge graph embeddings with domain-specific reasoning derived from experts or entailment regimes like OWL2. By loosely coupling these two techniques, users can not only enhance prediction accuracy by incorporating domain knowledge that may not be explicitly included in the input knowledge graph but also integrate their own knowledge graph embedding and reasoning methods.
Initial Results
The initial results demonstrate significant improvements in mean reciprocal rank (MRR) accuracy compared to vanilla knowledge graph embeddings. Researchers achieved up to 3 times enhancement in MRR accuracy and outperformed hybrid solutions that combine KGEs with rule mining and reasoning by up to 3.5 times MRR.
Evaluation Methodology
In order to validate this proposal, evaluations were conducted using both synthetic and real-world datasets including DBPedia (a real-world KG accompanied by an ontology) and LUBM (a popular synthetic KG generator). Due to computational limitations, researchers worked with a subset of DBPedia containing a well-connected subgraph consisting of the 12 most popular relations and related information. The models used for comparison include TransE, DistMult, ComplEx, HolE, ConvE, KALE, and plogicNet; all models utilized consistent hyperparameter settings for both vanilla KGEs (knowledge graph embeddings) and our approach during evaluation process. The hardware used for experimentation consisted of a machine with an AMD Opteron processor running at 32x2.3 GHz , 62GB RAM ,and an NVIDIA GPU .
Conclusion
Overall , this novel approach combining data - driven power of Knowledge Graph Embeddings with domain - specific reasoning has demonstrated significant improvements in mean reciprocal rank (MRR) accuracy compared to vanilla Knowledge Graph Embeddings . This research provides promising results towards improving the accuracy of Knowledge Graph Completion tasks which are essential for critical domains like healthcare .
