IDAMS News - Issue No 19 September 1998
WebWorldCommunication and MediaPublic Domain InformationBuilding the Infostructure
Electronic Networks
Virtual Learning Communities
Information Management Tools
Training in Information
 

 

 

  UNESCO CII SectorSearchSitemap

 

 
     IDAMS available in French and Spanish
     UNESCO International Training Seminar in 1998
     ICSOPRU Data Base available on diskettes
     Description of IDAMS programs and facilities

    IDAMS available in French and Spanish
 
Now the IDAMS software "speaks" also French and Spanish!!! Both the User Interface, including Graphid, and printed results of program execution can appear in English, French or Spanish. The User Manual is also available in these three languages. French and/or Spanish versions can be added when the English one is already installed.

 

 
    UNESCO International Training Seminar in 1998
 
As every year, CII/INF is organizing International Training Seminar: Introduction to IDAMS. The basic part of this Seminar will take place from Monday 23 through Wednesday 25 November 1998 at UNESCO Headquarters, Paris. Its purpose is to introduce the latest version of the IDAMS software package. Participants willing to extend their knowledge and/or perform more exercises could continue individual work on 26 and 27 November. The IDAMS development staff and computer rooms will remain at their disposal during this period.

English will be the main language of this Seminar although explanations in French can also be provided. Training material will be available in both English and French.

There is no registration fee, but all other costs (travel, board, lodging) are at the charge of participants.

 

 
    ICSOPRU Data Base available on diskettes
 
ICSOPRU is the acronym of an UNESCO international research project on the management, effectiveness and productivity of research teams and institutions to which they belong, conducted by the Organization between 1971 and 1989. Seventeen countries took part in this project, namely:
    • in Africa: Ghana, Nigeria
    • in the Arab States: Egypt
    • in Asia: China, India, Republic of Korea
    • in Europe: Austria, Belgium, Finland, Hungary, Poland, Spain, Sweden, Ukraine
    • in Latin America: Argentina, Brazil, Mexico.

An approach from multiple perspective was adopted as theoretical foundation of the project, taking advantage of recent developments in a series of domains such as system analysis, organizational psychology, sociology of sciences, managerial sciences. From the methodological point of view, it was decided to collect - through direct interviews with research teams' heads and members, and with institutions' heads – opinions and facts on a number of factors supposed to govern their scientific productivity and to influence the impact of their work. The same questionnaires, carefully translated into national languages, were administered in all countries participating in the same "round" of the study.

The common methodology (internationally developed for the ICSOPRU) was based on standard procedures: hypothesis formulation, construction of measuring instruments (i.e. questionnaires), sampling design, collection of data in standardized ways from large and heterogeneous population of R&D units, data verification and correction, construction of standard data files, etc. These ICSOPRU procedures constituted a technical Guidebook, constantly kept updated.

The international comparability of the study results mainly from the respect on the part of each country of these procedures especially as regards:

    • the use of a set of questionnaires internationally developed for the study
    • the administration of the questionnaires using similar interview techniques
    • the verification and correction of the data, and the construction of computer files using the same techniques and programs.

Taking into account the scientific interest that the ICSOPRU data may present for other decades, an effort was made to deposit international ICSOPRU data and relevant documentation at UNESCO Archives, and in a number of external institutions. The complete set of ICSOPRU data prepared for archiving in 1990 consisted of four tapes containing international data files in IDAMS and SPSS formats. Because of continuous development of microcomputers, the international ICSOPRU data files in IDAMS format were transferred to more popular media and are now available on diskettes.

ICSOPRU data and corresponding documentation can be obtained form UNESCO Archives, upon request addressed to:

Mr J. Boel
UNESCO, Archives Records Management & Micro Division
7, Place de Fontenoy
F 75352 PARIS 07 SP, France
Fax: (33-1) 45.68.56.17
Internet e-mail: j.boel@unesco.org

 

 
    Description of IDAMS programs and facilities
 
CLUSFIND (finding groups in data)

This program was initially developed by L. Kaufman and P.J. Rousseeuw at Center for Statistics, Vrije Universiteit Brussel, Belgium, and adapted to IDAMS using the source from the MICROSIRIS software.

CLUSFIND performs cluster analysis of objects (cases in an IDAMS dataset or row/column elements in an IDAMS square matrix) using one of six algorithms. Four of these algorithms perform their actual analysis on a dissimilarity matrix. Such a matrix can be input directly, or is calculated by the program if a dataset, a similarity or a correlation matrix is input. With a dataset as input, Euclidean or city block distance can be used for computing dissimilarities.

PAM (Partitioning Around Medoids) searches for k representative objects (medoids) which are centrally located in their clusters. The medoid is the object for which the average dissimilarity to all the objects in the cluster is minimal. Actually, the algorithm minimizes the sum of dissimilarities. The selection of k medoids is performed in two phases. In the first phase, an initial clustering is obtained by the successive selection of medoids until k objects have been found. In the second phase, an attempt is made to improve the set of medoids.

CLARA (Clustering LARge Applications) is also based on the search for k medoids, but it is designed for analyzing large data sets. Thus, the input to CLARA has to be a dataset. Internally, CLARA carries out two steps. First, a sample is drawn from the set of objects, and divided into k clusters using the same algorithm as in PAM. Then, each object not belonging to the sample is assigned to the nearest among the k medoids. The quality of this clustering is defined as the average distance between each object and its representative object. Five samples are drawn and clustered in turn, and the one is selected for which the lowest average distance was obtained. The retained clustering of the entire data set is then analyzed further.

FANNY (Fuzzy ANalYsis) is a generalization of partitioning, but the algorithm, instead of assigning an object to one particular cluster, gives its value of membership coefficient to each cluster, and thus provides much more detailed information on the structure of the data.

AGNES (AGglomerative NESting) constructs a tree-like hierarchy, starting with clusters of one object each, and proceeding by successive fusions until a single cluster is obtained with all the objects. In the first step, the two closest objects are joined to constitute a cluster with two objects, whereas the other clusters have only one member. In each succeeding step, the two closest clusters are merged.

DIANA (DIvisive ANAlysis) produces similar output to AGNES but constructs its hierarchy in the opposite direction, starting with one large cluster containing all objects. At each step, it splits up a cluster into two smaller ones, until all clusters contain only a single element. In the first step, the data are split into two clusters by making use of dissimilarities. In each subsequent step, the cluster with the largest diameter is split in the same way.

MONA (MONothetic Analysis) is intended for data consisting exclusively of binary (dichotomic) variables. The algorithm does not use dissimilarities between objects but directly the values of variables. At each step, one of the variables (x) is used to split the data by separating the objects for which x=1 from those for which x=0. In the next step, each cluster obtained in the previous step is split further, using values of one of the remaining variables. For each split, the variable most strongly associated with the other variables is chosen.

 
     IDAMS Newsletter Archives

 

Back to IDAMS menu

 

 


cii.webmaster@unesco.org
© Copyright 1998 - UNESCO