Within complex scienti c domains such as pharmacology, operational equivalence between two concepts is often context-, user- and task-speci c. Existing Linked Data integration procedures and equivalence services do not take the context and task of the user into account. We present a vision for enabling users to control the notion of operational equivalence by applying scienti c lenses over Linked Data. The scienti c lenses vary the links that are activated between the datasets which a ects the data returned to the user.
Developing an integrated view over pharmacological data is challenging due to the complexity of the domain. For example, searches for the chemical \Fluvastatin" on ChemSpider5 and DrugBank6 return di erent compounds: although their basic chemical structure matches, the compounds di er in their stereochemistry7. Another challenge is that the domain scientists have di ering opinions on when data records can be related. For example, when searching for information about \Protein Kinase C Alpha", the scientist may want information returned for that protein as it exists in humans8, mice9 or both. Additionally, when connecting across databases one may want to treat these proteins as equal in order to bring back all possible information whereas in other cases (e.g. when the researcher is focused on humans) only information on the particular protein should be retrieved.
Data integration in Linked Data relies on equality links between resources
across di erent datasets. Relevant existing approaches include Bio2RDF [
Chem2Bio2RDF [
We aim to support users in controlling and varying their view of the data by applying a scienti c lens which govern the notions of equivalence applied to the data. Users will be able to change their lens based on the task and role they are performing rather than having one xed lens. To support this requirement, we propose an approach that applies context dependent sets of equality links. These links are stored in a stand-o fashion so that they are not intermingled with the datasets. This allows for multiple, context-dependent, linksets that can evolve without impact on the underlying datasets and support di ering opinions on the relationships between data instances. This exibility is in contrast to both Linked Data and traditional data integration approaches. We look at the role personae can play in guiding the nature of relationships between the data resources and the desired a ects of applying scienti c lenses over Linked Data. 2
We present two personae representative of the scienti c users of the data
integration system being developed in the Open PHACTS project10 [
Chris the Integrative Systems Biologist. Chris is a researcher in a university who investigates processes in mice and compares these with processes in humans. When searching for data he needs to be able to switch between species depending on the kind of experimental (protein and mRNA) data he wants to compare in pathways, i.e. sometimes he will be interested in just mice data, sometimes just human, and at other times the combination of both. However, he always insists that genes and proteins are kept distinct. When evaluating toxicological e ects of Fluvastatin he is interested in both the pharmacological active stereoisomer of Fluvastatin and inactive one, since the latter still might have side e ects. He therefore allows the link between ChemSpider and DrugBank for Fluvastatin.
Fiona the Bioinformatician. Fiona works for a pharmaceutical company. She searches for potential drug-like compounds by mining existing literature for similar compounds to interact with known targets. For compounds that have been well studied, she would like to apply strict operational equivalence conditions, i.e. 10 http://www.openphacts.org/ accessed 7 Sept 2012.
Scienti c Lenses over Linked Data equating chemical compounds only if their stereochemistry matches and keeping genes and proteins distinct, in order to work with the most relevant data. However, for compounds that have not been so widely studied, she is willing to apply more permissive notions of equivalence to increase the volume of data returned. 3
Within scienti c datasets it is common to nd links to the \equivalent" record in another dataset. However, there is no declaration of the form of the relationship. There is a great deal of variation in the notion of equivalence implied by the links both within a dataset's usage and particularly across datasets, which degrades the quality of the data. The scienti c user personae have very di erent needs about the notion of equivalence that should be applied between datasets. The users need a simple mechanism by which they can change the operational equivalence applied between datasets. We propose the use of scienti c lenses.
A scienti c lens is de ned in terms of the scienti c notions upon which
operational equivalence can vary, e.g. stereochemistry, gene and protein equivalence,
and cross species matching. Our goal is that the scientists should be able to
focus on their science and not on the links between datasets. We do not propose
to de ne an ontology of relationships; Halpin et al [
Consider again the scienti c personae. Chris would have two lenses that he could switch between: one that keeps species independent of each other and one that matches proteins across species. Fiona would also have two lenses: one that applies strict levels of operational equivalence and a more relaxed one. She could of course de ne intermediary ones to allow her to slide the level of strictness. 4
Equivalence of di erent, but for some purposes comparable content from, data resources depends upon a user's opinion, task, and context. We have proposed applying scienti c lenses which vary the notion of operational equivalence between datasets. A scienti c lens varies the links that are active. Thus, a query over the Linked Data would see a di erence in the data returned when applying lenses with di erent notions of operational equivalence. Note that although we are developing our lenses in the context of pharmacology, the same principles could be applied in other domains. Our ongoing work is concentrating on two strands. First we are developing mechanisms to de ne scienti c lenses in terms of scienti c notions which can then be used to control which links are active. Second we are developing a query infrastructure to support the contextualised stand-o mappings used by our scienti c lenses approach.
Acknowledgements The research leading to these results has received support from the Innovative Medicines Initiative Joint Undertaking under grant agreement number 115191, resources of which are composed of nancial contribution from the European Union's Seventh Framework Programme (FP7/2007- 2013) and EFPIA companies' in kind contribution.