Ontologies. To clear up any possible confusion, when ‘ontology’ is mentioned in this disserta- tion, this does not refer to the branch of philosophy, but the information science concept: a representation of concepts in a specific domain (Xxxxxx, 1995). This is similar to a thesaurus or word list, with the most well known Dutch example being the Archeologisch Basisregister (ABR) ontology (Xxxxxx et al., 1992). For NER, it is useful to have ontologies for the categories of entities you are targeting, as whether or not a token occurs in such a word list is an indication that it might be an entity. For Artefacts and Time Periods, we use the aforementioned ABR ontology. This is a hierarchical list of artefacts, time periods and monument types, created and maintained by the RCE. We have slightly adjusted some of the entries to better match natural language, e.g. changing “bijl, doorboord” (axe, perforated) to “doorboorde bijl” (perforated axe). Unfortunately, the ABR is not very exhaustive and only contains a basic list of time periods. This is why we decided to use the PeriodO time appellations list (Xxxxxxxxxx et al., 2016) for translating Time Periods to year ranges (further described in chapter 7). We also altered this list by adding more time periods, mainly geological time spans (e.g., Holocene) and specific cultures (e.g., Bell Beaker Culture). For Locations and Species, we are not using any ontologies, as we are focusing more on Artefacts and Time Periods for the time being. For future work on these
Ontologies. 5.1.1. Symptom Ontology (SO) The Symptom Ontology (SO) [41] was designed around the guiding concept of a symptom being: "A perceived change in function, sensation or appearance reported by a patient indicative of a disease". The Symptom Ontology captures and documents the semantics of two sets of terms, the term “Sign” and the term “Symptom”. The ontology is open source. It was developed as part of Gemina project, starting in 2005 at the Institute Genome Sciences (IGS) at the University of Maryland. Work ended on 2009. In July 2008 the Symptoms Ontology was submitted for inclusion and review to the OBO Foundry and today the standardization body for the Symptom Ontology is the OBO Foundry. The ontology also provides human readable text together with computer readable format. The available computer readable formats of the ontology are in OBO and OWL. The semantics of the Ontology are coherent, consistent and there is a rigid domain specification. Last but not least the symptom ontology reaches high level of interoperability.
Ontologies. For the terminology used in the WP6 literature database no existing ontology was used. For the column headers (field names) the WP6 NanoReg terminology was mapped against existing ontologies. 258 terms in the Wp6 NanoReg database needed to be mapped to Uniform Resource Identifiers (URIs). 207 terms were mapped (~80%) but most of the terms were mapped against multiple URIs. For example dispersion protocol was mapped against the Nanoparticle ontology term 1969, dispersion, and the Information Artifact Ontology term 0000301, protocol (NPO_1969 IAO:0000301). 51 terms could not be mapped to URIs at all(~20%). This mapping was done only based on the exact term (not the actual meaning of the term). In future projects it is essential that mapping of ontologies and database fields it done on the meaning and interpretation of the term and field rather than a textual match. For example: “number of different incubation times” was matched to the terms “number”(PATO:0001555), “different”(C46003), “incubation” (npo:NPO_2000) and “times” (PATO:0000165). None of these terms in itself cover the meaning of the term “number of different incubation times”. The use of existing ontologies for the column headers helps in harmonizing the interpretation of the database fields and the interpretation of the database structure. This facilities for example data analysis and data entry. In the data-entry tool users are allowed to enter free text fields for various entries. Moreover users can extend and edit pick-list that are used for various fields. By doing this there has been no 1xxxx://xxx.xxxx.xxx/apps/group_public/download.php/30992/Money,%20C.,%20Reg%20Toxicol%20Pharma col,%202013.pdf control on the terminology that was used to explain various parameters. For example titanium dioxide can be explained by: Titanium dioxide, TiO2, tio2 etc. these are 3 terms meaning the same thing. An important lesson is that data entry should be restricted to specific terminology, preferably linked to a specific existing ontology as much as possible. This guarantees the description of parameters to be both unique and exclusive By doing this the interpretation of a specific term by different experts is fixed. Existing ontologies can and should be used for: Chemical abbreviations, nanoparticle characterization, cell assays, biological assays, physical conditions, physical chemical characteristics, media etc. Through the eNanoMapper database existing ontologies are used for the NanoReg2 and Calibrate ...
