ǂThe ǂenhancement of open data usability by defining a categorization method based on the open data portal metadata

[M. B. Frtunić Gligorijević]

114 lista

Biografija autora: list [160]. Bibliografija: listovi 149-159.

Electrical and Computer Engineering (Computer Science). Open data categorization.

Due to numerous data transparency and open government initiatives, a large volume of data was published on open data portals. To make it more accessible and visible, these portals have introduced data filtering by category, tags, format, organization, etc. This information is stored as metadata and provided when publishing the data. However, the metadata is not always complete. The lack of data categories has a great impact on the data visibility, accessibility, and usability of information. As the data increases on the portals, it becomes harder to find and identify the wanted information when the category is missing. Within this doctoral dissertation, an analysis of metadata on open data portals, as well as an analysis of categories and tags usage, and their connections on open data portals was performed. Afterward, the problem of missing data categories was addressed by proposing a methodology for data categorization based on the combination of tags. Within the methodology, the hierarchical organization of tags in a category was defined based on their usage in categorized data. Then, a tool was presented for visual analysis of the hierarchical organization of tags, and a proposal was given for the data categorization based on the combination of tags. The presented categorization relies on the way tags are used in categorized data, i.e. their hierarchical organization. The approach calculates the similarity between two tags, and two combinations of tags, as well as defines the parameters for categorizing the combination of tags with categories on the portal. Afterward, an algorithm was defined that proposes the categories for a dataset with a given combination of tags. For the proposed categorization, an evaluation was performed using the data from the Canadian open data portal. Lastly, within the doctoral dissertation, a model was proposed for supplementing the datasets’ metadata on open data portals.

otvoreni podaci, kategorizacija otvorenih podataka, analiza formalnih koncepata, kategorizacija

open data, formal concept analysis, open data categorization, cetegorization

(3:001):004.738.5(043.3)

T 120

Tekst