OSSOBOOK: Database and Knowledgemanagement Techniques for Archaeozoology 1
1
2
2,3
Hans-Peter Kriegel , Peer Kröger , Henriette Obermaier , Joris Peters , Matthias 1 3 Renz , Christiaan Hendrikus van der Meijden 1
Institute for Informatics Ludwig-Maximilians Universität Munich, Germany
{kriegel,kroegerp,renz}@dbs.ifi.lmu.de 2
Library for Anthropology and Archaeozoology Munich, Germany
[email protected] 3
Faculty of Veterinary Medicine Ludwig-Maximilians Universität Munich, Germany
{Peters,v.d.Meijden}@it.vetmed.uni-muenchen.de ABSTRACT This demo describes the OSSOBOOK database system developed for archaeozoology applications providing data storage, data retrieval, and data mining facilities. It shows a case study of integrating state-of-the-art database concepts like intermittently synchronized database system, as well as concepts of information retrieval and knowledge representation like similarity search and data mining in order to provide a comprehensive system for an interesting application domain.
1.
INTRODUCTION
Archaeozoology deals with the analysis of animal remains that have been uncovered by archaeological excavations to proof scientific assumptions and derive new hypothesis. For that purpose, researchers maintain archaeological excavations all over the world producing a tremendous amount of information that can no longer be analyzed by any human manually. Rather, the information on individual findings, e.g. descriptions of osseous parts, should be stored in a database system and should be available for analysis purposes like information retrieval and data mining via a central portal. Since the findings can usually not be transferred to other locations due to regional legal restrictions or high transportation costs, the information on these findings need to be recorded on-site. In order to avoid a redundant and error-prone data recording including a manual and ad hoc on-site documentation at the location of each excavation followed by a successive input of this documentation into a cen-
Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Copyright 200X ACM X-XXXXX-XX-X/XX/XX ...$10.00.
tral database at some institute, an advanced and integrated database system is envisioned. Furthermore, it should be capable of dealing with the application specific obstacles, in particular with the fact, that internet access may not be available during long field sessions all the time. Thus, the system must provide dedicated synchronization concepts in order to allow online access to the central database even if internet access cannot be guaranteed persistently at the location of a given excavation. Last but not least, the system should also be capable to share data from different institutions all over the world that work in different projects. Therefore, the system should enable an advanced project management. In this demonstration we present OSSOBOOK, a system that provides a solution for such archaeozoology applications. In particular, OSSOBOOK implements a mobile database system using the concept of intermittently synchronized databases [2] to handle the application specific problem that users may have no persistent access to the central system due to network constraints but may still want to update certain parts of the database. A client-server architecture ensures that each client (usually a particular researcher equipped with a laptop) manages its own local database that is schema-equivalent to the central database at the server. This way, each client can make its updates locally, independently, and—most important—offline. At a given time, e.g. when a network connection to the server is established, the client and the server synchronize, i.e. the updates of the client are inserted into the central database at the server and the updates of the server (e.g. originating from other clients that have already synchronized) are send to the client. The application scenario of the OSSOBOOK system is visualized in Figure 1. The central database is located at some institute. Within the institute, clients can connect to this central database system. On the field, each client has its own local copy of the central database and can connect from time to time to the central repository via a network for synchronization.
Field
Field
Field
Institute
Central DB WWW
Figure 1: The architecture of the mobile OSSOBOOK database system.
2.
THE OSSOBOOK DATABASE
3D models of bones or skeletons, and time series of findings. The data mining unit of OSSOBOOK provides with algorithms for standard data mining tasks such as clustering, outlier detection, classification and association rule mining to do specialized, domain-driven analysis of the entire database. For example, association rule mining based on the Apriori algorithm [1] is used for rating archaeological excavations. Every excavation can be treated as a market basket, the items are the found objects. From the application it is now interesting to find rules like “If item A has been found, item B will also be found with a probability of x%”. This gives the local excavation manager a list of items to look for.
The OSSOBOOK database is implemented as a MySQL DB with a JAVA GUI. For the purpose of collaboration between different institutes in different countries all over the world, the central database has a user management, which allows the institutes to manage there own projects independently from each other. Research results stay related to the institute. Each cooperating institute can maintain its own database, which will synchronize with the central database. Within an institute, clients connect either directly to a project on the local institute database, or they log in to the related project on the central database. In case of field work the researcher has a copy of the project on a local installed OSSOBOOK database on his notebook. He therefore can use and alter all the project related data on the field. As soon as the researcher has internet access he can synchronize his local database with the institute database, which then will synchronize with the central database. On the central database all the research data is stored in the different projects of the different institutes. Here the data mining algorithms are provided to all participants. These data mining concepts use the complete dataset of all collaborating institutes. This is the main scientific advantage of the OSSOBOOK database. Similarities in data collected in different parts of the world, on different time periods, by different institutes can be shown, independently from the internal use of the data. OSSOBOOK is an intermittently synchronized database system which operates in double depth. First stage is the synchronization of the local databases of the field clients to the local institutes database, the second stage is the synchronization of the institute database to the central database. Within these stages data input can be done at the field clients and at the local institute clients. The data mining concepts can be used from the local institute clients at the central database without direct access to the under laying central data. On all stages the requirements of an Intermittently Synchronized Database System (replication, synchronization, error tolerance) are met.
4.
3.
7.
RETRIEVING AND MINING DATA WITH THE OSSOBOOK SYSTEM
A CASE STUDY
The Institute for Palaeoanatomy of the Ludwig-MaximiliansUniversity (LUM) Munich, Germany, participates at an archaeological excavation in G¨ obekli Tepe, Turkey. About 30.000 bones of different animals are analyzed. Originally, these data have been recorded manually. Now the OSSOBOOK system is used for data input and data storage. In this demo, we will exemplarily see how archeological data from the G¨ obekli excavation is input into a local client and then synchronized with the central database at LMU. The data consists of classical relational data as well as complex data like pictures and 3D spatial objects. We will also show how the data mining capabilities of OSSOBOOK give a rating of the current findings in the site taking other projects of the LMU all over the world into account. In particular, OSSOBOOK can derive a list of items each associated with a probability that it can still be found in G¨ obekli Tepe.
5.
AVAILABILITY
The OSSOBOOK system is accessible at http://www. dbs.ifi.lmu.de/,
6.
CONCLUSION
first-hand experiences with research prototype systems In this demonstration we present the OSSOBOOK system for archaeozoology applications that integrates core DB, KM, and IR concepts like and provides first-hand experiences with research prototype systems from these areas. We show how the synchronization of multiple users in a clientserver environment can be performed using the concept of intermittently synchronized database systems. Furthermore, we demonstrate how specialized data mining techniques can provide important application specific services like the rating of archaeological excavations, etc. Last but not least, this demo outlines a completely new application area that provides a variety of novel problems for researchers in the areas of DB, KM, and IR.
REFERENCES
[1] R. Agrawal and R. Srikant. Fast algorithms for mining association rules. In Proc. SIGMOD, 1994. OSSOBOOK not only provides a standard relational database [2] S. Mahajan, M. Donahoo, S. Navathe, M. Ammar, and system for storing data of archaeological findings. It also S. Malik. Grouping techniques for update propagation provides facilities for retrieving and mining this data. in intermittently connected databases. In Proc. ICDE, Beside the standard retrieval capabilities of relational database 1998. systems, OSSOBOOK provides the possibility to perform similarity queries on more complex objects like 2D pictures,
