Name
Abstract | ||
|---|---|---|
There are many connections between graph mining and inductive logic program- ming (ILP), or more generally relational learning. Up till now these connections have mostly been described or exploited in an informal way. A more careful and formal study of them may lead to new insights in the relationships between the dieren t learning formalisms, and possibly new methods for solving certain problems that rely on translating problems or algorithms from one setting to another. In this paper we initiate an investigation into this. It has been observed many times that there are connections between graph mining and inductive logic programming (ILP), or more generally relational learning. So far, however, these connec- tions have mainly been described in an informal way. Moreover, there are many possible ways of translating data / hypotheses from one representation framework to another. A more careful study of such translations may give us additional insights in the relationship between dieren t settings, and could directly lead to two ways of improving the current state of the art in graph mining. The rst is \translation of algorithms": algorithmic techniques could be transferred from one framework to the other. The second is \translation of data": one might be able to solve problems by translating the data from one framework into the other, running a standard data mining algorithm for that framework, and then translating the results back to a proper solution for the original data. Several authors have already exploited similar connections (e.g., (5, 4)), but it seems that many more connections could be investigated. In this paper we have a brief and non-exhaustive look at a number of connections between graphs, hypergraphs, and rst order logic representations, and formulate a number of questions that this leads to. We hope that this discussion may be inspiring with respect to the development of new algorithms or unexpected application possibilities for existing algorithms. |
Year | Venue | Keywords |
|---|---|---|
2007 | MLG | relational learning |
Field | DocType | Citations |
Functional logic programming,Inductive logic programming,Horn clause,Computer science,Inductive programming,Multimodal logic,Theoretical computer science,Prolog,Artificial intelligence,Logic programming,Dynamic logic (modal logic),Machine learning | Conference | 2 |
PageRank | References | Authors |
0.38 | 5 | 3 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Hendrik Blockeel | 1 | 2744 | 177.48 |
| Tijn Witsenburg | 2 | 17 | 1.51 |
| Joost N. Kok | 3 | 1429 | 121.49 |