A System for Referencing Personal Names through Iconography and Sharing an Authoritative Information Source for Personal Names by API

A System for Referencing Personal Names through Iconography and Sharing an Authoritative Information Source for Personal Names by API
Togiya, Norio, University of Tokyo, togiya.norio@iii.u-tokyo.ac.jp
Kawashima, Takanori, Tokyo Institute of Technology, t_kawa@valdes.titech.ac.jp
Abstract
In this research, we constructed an ontology-based name authority file using a topic map, and then, using API and other Web services, we looked into using this with navigational systems .The name authority file was created and provided through API services. By using this service, it is possible for developing navigation and analyzing tool of Digital Cultural Heritages to have a comprehensive name list of historic personal names and artists. These API are used for Digital Cultural Heritages of digitized anthropological material, old photographs.

Introduction
While there have heretofore been various forms of Digital Cultural Heritage, even for digitized and stored materials it has been typical to select materials through a list or by alphabetical search. However, in situations such as for historical documents, it is preferable to be able to select materials by searching for documents that deal with related individuals, social organizations, historical events or periods. In particular, a proper noun should be controlled as authority file.

For this reason, in this study, an ontology-based name authority file —which defines the relationship between people, organizations and places—was created. And the information was stored in standardized ontology language. We adopted Topic Maps for constructing name authority file and API system providing it for digital cultural heritage.

There are some examples of ontology-based name authority file using Topic Map. The unique features of this study is visualizing ontology-based name authority file using Topic Map and providing it with API. In this paper, we will describe in detail.

Definition of Ontology
An important goal of this archive is to select and view each material through an understanding of the various materials and the relationship between the characteristics related to them. To achieve this, we decided to use ontology relating the various items in this project:

Defining concepts, characteristics, and meanings of various items
Aiming at systematizing various concepts and items in the world
Aiming at generality, which makes materials reusable, and shared knowledge
Describing items with rules and in a language based on certain regulations
In Greek, ontology originally means existence and has been frequently used in philosophy. However, with the development of the study of artificial intelligence and knowledge engineering since the late 20th century, it began to be used as a term referring to a semantic systematization method for various items in the world in order to facilitate its understanding by machines. Meanwhile, when the Semantic Web was proposed by Tim Berners-Lee as a more effective method to connect fragments of information on the Web through the spread of the Internet in the late 1990s, ontology was used as a term to refer to the methodology used to describe content created on the Web with more regulated rules and language.

Various types of ontologies have been developed through the course of time, such as ―Upper ontology aiming at the systematization of various items in the world based on philosophical discussions, Domain ontology developed to express the edifice of knowledge in a specific domain (mainly for industrial use), and Web ontology developed to systematize information and knowledge on the Web.

“Upper ontology and Domain ontology are provided to construct concepts 1), 3), and 4) aiming at 2) defined above. Web ontology initially had the practical purpose to develop item 4) in order to define the relationship between information items on creating contents using method 1) to achieve item 3), which can be seen in the combination of the Dublin Core and RDF technologies. The ontology of CIDOC CRM –used to describe the metadata of cultural resources–seems to belong to this type. Recent advances made in Web ontology aim at the semantic systematization of content on the Web, which was its initial objective. In this context, it may be said that it has advanced in the area of item 2) by replacing the world with the Web. However, how the system of the concepts on the Web and that of the things in the real world are unified as the former is a reflection on the latter remains to be seen.

This study systematizes item 2) for historical information with method 1) in anticipation of the union of these in the future and aims at item 3) by using item 4), which can be unified with Web ontology, etc. in the future. As for now, we are mainly focusing on implementing items 1) and 2). In particular, we designed name authority file based on the above mentioned definition on ontology.

Constructing Ontology Using Topic Map
Introduction about Topic Map
We adopted topic map to constructing above mentioned ontology-based name authority file. Topic maps that represent ontology use the ISO/IEC JTC1 SC34 established ISO standard, which is still being revised, but as of April 2009 some parts have not yet received JIS standardization. Topic maps are a technique for classifying, organizing and making easily visible information and knowledge, and play a role similar to the index of a book for displaying information in space. It is possible to model and process the relations between subjects in the problematic areas, relations between information resources, such as Topics, Association and Occurrence with a computer.

In order to identify and discern the subjects, topic maps have a mechanism known as PSI (Published Subject Identifier). PSI, as a subject identifier, allocates and publicizes a unique IRI (Internationalized Resource Identifier) to each subject. Placed above the indicated address of the IRI, and serving as a descriptive information resource is the Published Subject Indicator, or PSD (Published Subject Descriptor), which allows the understanding of what the subject represents. For this project, as PSI was used, subjects were not given their own name or alias; rather they could be identified and discerned by their IRI.

Creation of Topic Map
Information relating to the photographer, scholar and noble family was collected to use navigation and analyzing for old photograph. Especially for information that needs to be shared, information necessary for multiple people is gathered. Information gathered in this manner is described in the creation of topic maps.

Information gathered by multiple people is shown in table 1 under the entries of "name," "reading," "nationality," "occupation," "birth and death," "biography," "hometown," "place of residence," "attached organization / group" and "title." These items are candidates for topics when creating the topic map. After completion of the topic map, navigation will become possible based on these subjects.

The three elements of which Topic maps are composed are mainly used to express a variety of subjects (topics), express relations between topics (Associations), and to link information resources related to topics (Occurrences).

These structures are adapted as shown in Table 1. The names, various related items, and reciprocal relations of items regarding the ontology at first can be thought of as topics for topic maps. Also, for name sources, each "person" and "place" and the relation among people (Associations) is adapted. Furthermore, as links to specific information resources, "birth and death," "biography," "related URL/URI" etc., correspond to appearance (Occurrences).

In addition, it is possible to classify information stored for these (Topics), (Associations) and (Occurrences). For the first topic (Topics) various "forms" were established; Topic Types, which represents the topic "form," for (Associations), defining the relation type are Association Types and Association Role Types, and for (Occurrence), Occurrence Types expresses the type of appearance.

The specific items of these "forms" can be expressed, for Topic Types, "Country," "Person," "place name," "organization/group," and "occupation" etc.

Furthermore, Association Types are established showing various relations such as "friendship relation," "marital relation," "student/teacher relation," "workplace relation", and "filial relation," Association Role Types express the role of each topic that produces relations, such as "significant role," "brother (sister) role," "wife role," "child role," and "pupil role," etc.

Next, Occurrence Types are established, "reading of name," "reading of surname," "name in roman alphabet," "surname in roman alphabet," "personal relationship," "relationship period," "birthplace," "source," "role," etc. Based on these items information is structured.

Personal topics such as parent-child relation, fraternal relation, matrimonial relation, friendship association etc. are expressed through the lines leading into and out of the personal topics.

After creating the topic map, based on these forms queries and navigation are possible, and compared to semantic processing more meaningful searches can be conducted.

Table 1. Major Components of the Topic Map

Categories of personal name authority information Correspondence in Topic Map
Name Topic Name
ID SubjectID
Related URL/URI External occurrence
Dates of birth and death Internal occurrence
Brief biography Internal occurrence
Place of birth (multiple responses possible) Linked by association to other topics
Place of residence (multiple responses possible) Linked by association to other topics
Relationship (multiple responses possible) Association
An Overview of Iconographic Analysis Using Authoritative Information.
In this study, as shown in Fig. 1, an ontology-based name authority file was constructed, and that information was distributed through API. For name authority file, focusing on photographers, scholars and noble family, roughly 3000 names were entered into a DB system. This authoritative personal name information was mainly made from standardized reference materials used in museums, libraries and archives, and created from element sets which allowed the sharing of information in this study, and API was added to the database to enable external searching. In addition, as an external system a picture annotator was created using this API.

Figure. 1. Mutual Relationships among Topics

Full Size Image

In this study, a simple API was designed to make personal information taken from a reference DB searchable. By using this API, it is possible to use even personal name information controlled by an external system.

There were four Parameters that could be used (personal name, birth year and year of death, family, personal ID) to search, and the eight types of data that could be obtained from search were (personal name, personal ID, birth year or year of death, family, biography, related individuals and their relations, and source). It was possible to identify even people and families that had a unique ID, as well as individuals who shared the same name. Furthermore, by using the controlling ID, semantic interchange with an independent external system could be made possible.

The goal of the current design is not to have all fields of the DB searchable by semantic search, but to provide those fields thought to be frequently utilized by external applications simply. If a higher order search is desired, it can be performed from the editing UI of the authoritative information DB. The XML-RPC was used for the API protocol.

Designing Name Autholity File for Navigation and Analysing Image
In the last section, we constructed an ontology-based name authority file using a topic map. It is possible to use these to confirm interpersonal relationships and relationships between people and organizations while looking at a graph. In this research, we constructed a name authority file concerning scholar, photographers, etc. with a topic map. Fig 1 is a graph showing these people’s interpersonal relationships and their relationships. This chart mainly shows the relationships between the parties in relation to the anthropology materials. In this way, it is possible to use a personal name information authority using topic maps for navigation systems.

Figure. 2. Relationship among photographers

Full Size Image

Fig 2 is a graph showing relationships of photographers. Also, as shown in Fig 3, it is possible to show detailed information regarding these personal names, etc. by clicking on the personal name icons.
Figure. 3. Name Authority File

Full Size Image

Furthermore, it is possible to assign annotations to the materials, as well as to assign personal name information which has been regulated by the topic maps. It is also possible to provide information to the annotations using the regulated personal name information. With this, it is possible to search through material annotation information from the personal name information that has been regulated by the topic maps.

It is possible to combine this topic mapped personal name information with the digital cultural heritage main body, but it is also possible to put it to use through API. By using API, it is possible to use this information for navigation and annotation with other digital cultural heritages. This makes it possible to regulate the information and analyze the materials using information which defines the relationships.

Results and Issues
As seen above, ontology-based name authority file utilizing topic map and API for Digital Cultural Heritage was explained through application. In this study, ontology-based name authority file was constructed. By using topic maps, related people, can be clarified, and materials can be viewed. And API accelerated utilization of name authority file

In addition, by allocating URI to personal topics, personal identification can be performed by URI, not just by name, and by using the merge feature it is possible to share data from other topic maps. Also, by publishing with PSI information can be shared by other RDF and topic maps. For IRI, which kinds of character strings are appropriate is a matter for future consideration.

Concerning visualization, as demonstrated in fig. 2, visualization of relationships is possible. However, this causes problems such as the difficulty in recognizing relations that appear between more than three topics to arise. This problem also occurs with other viewers, and further study is necessary.

It is also necessary to consider whether and how to share the topic map created for this project with other projects and databases. As mentioned earlier, as each topic is given a universal URI, it is necessary to consider how it is necessary to link to other topic maps.

Furthermore, for this project, previously created information was converted into a topic map, but in the future it will be necessary to design a system to handle increased topics for Digital Cultural Heritage. Along with the need to store more materials, a system to increase information is necessary.

References:
ACKNOWLEDGMENT
I would like to give special thanks to Motomu Naito (Knowledge Synergy Inc), Hirotada Kobayashi (Cooba Corp).

(link)

Kivelä, A. Lyytinen, O. “Case study: publishing large collection of artworks using Topic Maps, ” Topic Maps Users Conference, Oslo, 2007

Norio TOGIYA Akira BABA “Constructing Integrated Digital Archive Using Ontology and User Community, ” Archives & Museum Informatics, ICHIM07 International Cultural Heritage Informatics Meeting Proceeding, (link) 2007

Riichiro Mizoguchi “Ontology Engineering Environments, ” Handbook on Ontologies, pp. 275-298 2004 Springer

(link)

Nicolas Guarino “Some organizing principles for a unified top-level ontology, ” Working Notes of AAAI Spring Symposium on Ontological Engineering, , Stanford, 1997