Literature mining method RaJoLink for uncovering relations between biomedical concepts

Abstract

To support biomedical experts in their knowledge discovery process, we have developed a literature mining method called RaJoLink for identification of relations between biomedical concepts in disconnected sets of articles. The method implements Swanson's ABC model approach for generating hypotheses in a new way. The main novelty is a semi-automated suggestion of candidates for agents a that might be logically connected with a given phenomenon c under investigation. The choice of candidates for a is based on rare terms identified in the literature on c. As rare terms are not part of the typical range of information, which describe the phenomenon under investigation, such information might be considered as unusual observations about the phenomenon c. If literatures on these rare terms have an interesting term in common, this joint term is declared as a candidate for a. Linking terms b between literature on a and literature on c are then searched for in the closed discovery to provide additional supportive evidence for uncovered connections. We have applied the method to the literature on autism and have used MEDLINE as a source of data. Expert evaluation has confirmed that the discovered relations might contribute to a better understanding of autism.

Discovery of Novel Term Associations in a Document Collection

“…Our approach shares some mental similarity with RaJoLink [8] even though it works in a different setting. Given a collection of articles on some topic, RaJoLink starts by finding rare terms in it.…”

Discovering Temporal Bisociations for Linking Concepts over Time

“…Approaches working on the first strategy have focused mainly on the automatic tools to select the intermediate B concept, for instance in [4] connections are extracted in the form of association rules and the possible intermediate concepts are identified with the integration of domain ontologies. For the second strategy, particular attention is paid to the pruning techniques which eliminate meaningless connections B ⇒ C. For instance, in [10] the idea is that of considering terms B with respect to a term-frequency based measure, named rarity, while in [11] the authors propose knowledge-based heuristics to provide a ranking of the connections B ⇒ C. Therefore, linking concepts over time seems to be a not yet investigated issue that would facilitate the discovery of connections between concepts only through a linking process over time. It is noteworthy that the problem here investigated is not different from the bisociation discovery seen in [1], [13], [9] where bisociations connect concepts from unconnected domains identified according to some view of data.…”

Bisociative Literature Mining by Ensemble Heuristics

“…In spite of MeSH filtering, the list of interesting terms can still be long and estimating the potential of a particular bridging term candidate to lead to useful bisociations is based on the expert's knowledge and intuition. The expert's involvement assures that the search is guided towards promising bridging concepts which are meaningful and interesting for the expert [11]. Therefore, we believe that experts' involvement should remain an important part of the process.…”

“…For example, in the RaJoLink methodology presented by Petrič et al [11] the list of interesting terms is effectively filtered according to MeSH (Medical Subject Headings) categories; in the next step the expert checks which of the remaining terms seem to be promising. In spite of MeSH filtering, the list of interesting terms can still be long and estimating the potential of a particular bridging term candidate to lead to useful bisociations is based on the expert's knowledge and intuition.…”