Skip to end of metadata
Go to start of metadata

Dominic Oldman, High Level, addition to tender specs
Largely based on Doerr's "New Framework"

ResearchSpace

High Level Brief for Semantic CRM search tool

Author: Dominic Oldman

Date: 14 th June 2011

Introduction

The semantic search tool is at the core of the ResearchSpace system and where all operations with ResearchSpace data begin. The design needs to provide the right level of precision and recall to support the location of particular records and to allow researchers to explore relationships between different datasets. The user interface needs to support powerful searching yet also be intuitive so that it can be used by a wide range of people, technical and non-technical.

This document is intended as a starting point for the UI design of the search tool which is included in the tender along with other treatments of the ResearchSpace UI environment. The requirements specified are not intended to be exhaustive but to provide a starting point for suppliers bidding for the UI design so that they can assess the assignment. Additional questions will be addressed at the bidder’s conference and can also be directed at the author.

The framework for this search tool is intended to provide a system of searching not available in traditional keyword search mechanisms and to transform the way in which scholars can work with harmonised datasets.        

Categories and Relationships

These search requirements are based on information and papers produced by Martin Doerr, Institute of Computer Science at F.O.R.T.H. In particular the idea that the search system should utilise a core set of CIDOC-CRM categories and relationships.

The data stored in ResearchSpace will be imported as RDF and mapped to the CIDOC-CRM ontology structure. The CRM ontology provides the basis and framework for the search mechanism. However, the extent of the categories and relationships that should be supported by the search tool is based on a framework proposed by Doerr and Tzompanaki [1] where they are called ‘Fundamental Categories’ and ‘Fundamental Relationships’. Suppliers may wish to familiarise themselves with the CIDOC-CRM ontology which is documented at www.cidoc-crm.org

These are reproduced as follows;

 

Domain

Range (query parameter)

 

Thing

Actor

Place

Event

Time

Thing

2.is part of

3.is similar or

the same with

4. has met

5. from

6. is origin of

8. refers to

9.is referred by

4.has met

5.from

8.refers to

9.is referred by

4.from

8.refers to

9.is referred

to at

 

4.from

8.refers to

4.from              

Actor

4.has met

6.is creator or

provider of

8. refers to

9.is referred by

 

2.is member of

4. has met

5.has parent or

founder

6.is parent or

founder of

8.refers to

9.is referred by

4.has met

5.from

8.refers to

9.is referred

to at

 

4.has met

8.refers to

6.has met

8.refers to

6.has met              

4.from              

 

Place

5.is origin of

8.refers to or is

about

9.is referred by

5.is origin of

8.refers  to  or

about

9.is referred by

2.is part of

5.is origin of

9.is  referred by

5.is origin of

7.at              

Event

5.is origin of

9.is referred by

8.refers to or is

about

5.is origin of

8.refers  to  or

about

9.is referred by

8.refers to or

is about

7.at

8.refers to

or is about

2.is part of

8.refers

or is about

7.at

Time

 

5.is origin of

5.is origin of

5.is origin of

5.is origin of

2.is part of

 

Concept

The category Concept plays a special role. Concepts can be subdivided into

subtypes of the FCs themselves, such as “Thing Concepts”, “Place Concepts”, etc.

 

·                   It should be possible to indicate the datasets (by namespace) to be included in a search either by choosing those that should be included or excluded. Tools should be available for making selection and clearing of selections straight forward. Users should be able to save their selections for other sessions. 

·                   The system should allow users to type a search term and for authorised terminology to be suggested as each letter of the search is typed, if the string is contained within a term. There is a possibility that synonym (and other helper features) functional could also be included.

·                   The system should be capable of being configured so that the system only draws upon a certain level of terminology (or all) in a hierarchy (e.g. configuration so that a search will only find potential terms at the first or second levels of a thesaurus with multiple levels of narrower terms). However, the search itself could be configured to use the narrower terms of the term selected.

·                   There should be an opportunity for the user to consult an authority list or thesauri hierarchy directly if they are unable to find the correct term through the suggestion system above.

·                   Search results should be categorised according to the fundamental categories above. (Results should belong to a class reflecting these categories either directly or through a sub class.)

·                   The number of results should be indicated against each category.

·                   Clicking on a result category should..... [what happens when a results category opens up]

·                   It should be possible to explore data by using the Fundamental Relationships between different categories. For example, a thing may have a relationship with an actor, which will relate to a place or period which may then relate to other things. The interface should provide the ability to explore in this flexible manner.

·                   The tool should be configurable to work with other CRM data implementations outside ResearchSpace. For example, it should be possible for the tool to be used on the British Museum’s Collection Online system (COL), if COL data complied with CIDOC-CRM and used RDF.

·                   The tool should be compatible with co-referencing resources where terminology is mapped across different organisations vocabularies.

·                   It must be possible to extract data and links from results data for use in other tools and environments. The ResearchSpace environment requires that results be used in other tools and that there be a data and information exchange with the ResearchSpace collaborative environment. (See main functional specification.)

·                   Traditional search mechanism should also be included and additional concepts for searching are discussed in the main ResearchSpace functional specification.

 


[1] A New Framework for Querying Semantic Networks

Katerina Tzompanaki1, Martin Doerr, Institute of Computer Science, F.O.R.T.H.- http://www.ics.forth.gr/tech-reports/2011/2011.TR419_Querying_Semantic_Networks.pdf

 

Labels:
None
Enter labels to add to this page:
Please wait 
Looking for a label? Just start typing.