The card sorting exercise was repeated here focused on search functionality
Oct 23, 2011
Hollie - These visualisations are very helpful in getting to the crux of it and subject to feasibility I would have the follwoing observations which relate to the Doerr / Tzompanaki article.
-The current wireframe doesn't really reflect the FC and FR concept and I have provided a summary of the Doerr article below. The concepts that may or may not be capable of incorporation are;
1. FC's (or the what, where, when, who concepts) are the main framework which are then qualified by the Fundemental realtionships. These provide the more semantic element (when compared to SQL statements and the queries based on SQL systems) and are backed up by controlled vocabularies (authority files and thesauri)
2. The comcept that the user can see what effect each additional FC to FR has on the results set. For example, in the current British Museum advanced search the user can select terms and then clicks the search button to see the results but then has to go back if they wish to refine the search. The Doerr model works on the basis that the user can directly see if a FC/FR combination has altered the amount of recall beyond the pouint that they wished and can be quickly rectified.
3. The ability to see exactly what you have queried is important but from the perspective of the FC and FR relationship.
4. We would need to see how these things related to external data as these have not been harmonised and therefore the searching logic and vocabularies may be a bit hit and miss unless we incorporated those specific vocabularies into RS (which could be done). However, may repositories will not have the relationships and therefore the quesry would have to be interprested differently.
5. I have uploaded an early fucntional spec again from the COFROM project that may illustrated better. It is[attached to this page.|^3DCOFORMfunctional.pdf]
-We probably still want to advance the idea of fundemental concepts and relationships as outlined in the Doerr and Tzompanaki . www.ics.forth.gr/.../2011.TR419_QueryingSemanticNetworks.pdf
This being Thing, Actor, Place, Event, and time thereby incorporating the classic When, What, Where and Who.
What you need to remeber is that there are two levels of interaction with the data. At the first level all datasets that come into the CRM have been organised according to a static framework. In other words they all share the same skeleton from which more specific information is then associated.
One of the key aspects of the COFORM system is that you can see the effect on the results set as you test and refine each facet.
The key elements from the article are these;
-Closed systems (i.e just representing one particular dataset) are able to have precise and accurate searching systems based because the data is consistent.
-Open systems have tended to have more keyword approaches which provide large results set but without accurarcy. However, advanced search options are unpopular because they cannot cope with different schema and languages.
-Semantic networks are open and therefore using many different kinds of proprties means that the amount of recall in the results set can be too small.
-The FC / FR system provides a middle way throught his to concerntrate on a fewer number of fundemental conepts and relationship. The smaller number of Concepts provides better recall and are more intuitive and can then be commbined with more advanced techniques if required. The FC's or 4 facets are based on the (4w) who, what, where, when.
The Finish culturesampo http://www.kulttuurisampo.fi/ used not 4 facets but 9 which it thought most relevant to cultural heritage including material, events, object types and for each facet uses a rich set of thesauri
Note that Doerr identifies another problem - "As last problem, the lack of recall in one field, due to parameter alternatives or
incompleteness of knowledge, roughly multiplies with the number of fields put in
conjunction. A typical query in culture may ask for a particular type of things from a
particular period of time, a particular geographic area, used/made by a particular
group of people. If each element of the conjunction has a recall of 90%, the
conjunction of four has about 66%, in the case of no correlation between the fields."
Doerr strives for a system that increases recall but as different facets are tested provides the user with an indication of the effect that that facet is having on the recall, as they go along. I think this explains the process in the 3D coform model.
He chooses 6 categories (and is looking at adding one more in the future dependent upon further work (dimension)that represent the 4w facets - and these cover 'a "Base level" distintion, but are nether completely disjoint nor absolute'. In other words they are the main starting point.
For Fundemental Relationship they have chosen ones which still have an intuitive nature but are not as ambiguos as say, 'from'. They choose relationships that provide the following.....
'(1) how and what something is (classification,
part-whole structure), (2) what an item has undergone gone in its history, and (3) what
it may “show”, say or refer to. We have not looked at relationships of intention,
motivation or cause, because they are rarely documented. In our current
Their selection is...
1. "has type: denotes relations of an item11 to a classification, category, type,
essential role or other unary property, such as a format, material, color. It
generalizes over dc:type, dc:classification, dc:format, dc:language. The
relationship is applicable to all FCs and has always range Concept.
2. is part of: denotes structural relations of an item to a wider unit it is contained
in. The relationship is applicable to all FCs, except for Concept. In case of
Actors, one would rather speak of “is member of”, and persons are the
minimal elements. Domain and range must be identical.
3. is similar or the same with: denotes the symmetric relation between items that
share features or are possibly identical. It is only usual for Things to
document similarity manually. There exist enough comparison algorithms that
deduce degrees of similarity automatically. We do not deal with these in this
4. has met: denotes the symmetric relation between items that were present in
the same event, including time intervals and places. Applicable to any
combination of FCs, except for Concepts.
5. from, has founder or has parent: denotes the relations of an item to
constituants of a context in its history which is either significant for the item,
or the item is significant for the context, “provenance” in the widest sense,
including time intervals and places. In case of genealogy or group formation,
natural language prefers the terms parent and founder respectively in order to
refer to Actors. The relationship is a special case of has met. The applicability
is analyzed in Table 1.
By “item” we mean any instance of a Fundamental Category, similar to “resource” in RDF
6. is origin of, founder of, parent of: the inverse of from, has founder or parent.
In case of Actor as domain, one would rather speak of “is provider or creator
7. at: denotes the relation of an Event to the Time and Place within which the
8. refers to: denotes the relation of an item that is information, contains
information or has produced information to the item this information refers to
or is about. The relation can even be extended to a Place from where such
9. is referred by/ is referred to at: the inverse of refers to.
I like the feature of explaining the scope note and it is the ambition of the Museum to provide representative images of authority terms in this way. many scope notes will not have associated images and autoseelction on images may be inaccurate.