The t-Protégé Project

The t-Protégé Project
Research area: 
Spatiotemporal databases
Semantic web & Ontologies

Introduction

In the past few years, ontologies have emerged as a valuable tool for representing the semantic context of a domain; ontologies provide the required formalism for computers to perform automated reasoning and the high level semantics for humans to work with. A number of tools have emerged that facilitate the creation, maintenance, usage and visualization of ontologies, such as Protégé, KAON, etc. Insofar, however, ontology tools are targeted to maintaining ontology snapshots, i.e. the most recent version of ontologies; past values of properties or relationships are not maintained, neither the period that the values/relationships were in effect can be stored. Some tools encompass ontology versions, which though cannot be effectively used for maintaining entity evolution since, generally, the number of changes in properties and/or relationships is too high, resulting in an excessive and unmanageable number of version.

This work is an extension of the Protégé tool to accommodate the modeling and presentation of entity history, i.e. past values of properties and/relationships; each such value is timestamped with the period that it was (or will be) valid in the real world. To this end, the presented extension provides:

  1. integrated support for data types expressing time quantities – more specifically dates (individual points in time) and periods (anchored segments of the time axis).
  2. data types for storing histories of properties of different types (strings, integers, floats, booleans and instances [i.e. relationships]).

The extension can be used in contexts that the modeling of entities’ history is important, such as historical archives, museums, etc.


Note: The wt-Protégé project is an extension of t-Protégé, providing additionally support for weighted data.

Downloading and Installing

For installing the extension you need to perform the following steps:

  1. Download Protégé Version 3.1.1 from the Protégé web site (http://protege.stanford.edu)
  2. Install the Protégé software following the instructions provided by the Protégé team
  3. Download the protege.zip file; this is the compiled, ready-to-run version of the extension.
  4. Replace the protege.jar file in the Protégé installation directory (in Windows-based systems, the default location is C:\Program Files\Protege_3.1).

You may now run Protégé as usual, and the extended version will be loaded.

You may also download the source code (original Protégé code plus the modifications and additions to implement the extension) from the following locations:

Available extensions

The extended version of Protégé encompasses support for:

  1. data types expressing time quantities – more specifically dates (individual points in time) and periods (anchored segments of the time axis).
  2. data types for storing histories of properties of different types (strings, integers, floats, booleans and instances [i.e. relationships]).

These extensions are presented in the following paragraphs.

The "Date" Data type

The "Date" data type provides support for expressing individual points in time. The precision with which the time point may be specified varies from year-level to the level of a second. Dates are given and displayed in the ISO-standard format, i.e. YYYY-MM-DD hh:mm:ss (YYYY = year in the range 0000-9999, MM = arithmetic month of the Gregorian calendar in the range 1-12, DD = day within the month in the range 1-31, hh = hour in the range 0-23, mm = minute in the range 0-59 and ss = second in the range 0-59). The following examples illustrate acceptable values for a date slot:

  1. 2006
  2. 2006-03
  3. 2006-03-25
  4. 2006-03-25 10
  5. 2006-03-25 10:58
  6. 2006-03-25 10:58:43

The precision used to denote the time quantity will be termed as granularity in this document. Note that the year is mandatory. Two more values are acceptable as dates, more specifically the strings unknown and now. The unknown value can be used to signify that the date is not known, while the now value should be interpreted as always being equal to the current "wall clock" indication.

The "Period" Data type

The "Period" data type provides support for expressing anchored segments of the time axis, i.e. time intervals with a specific beginning and a specific end. A period is entered as a pair of dates, enclosed in square brackets ([]) and separated with a comma. The dates may be expressed in different granularities, but it is required that the starting date should be less or equal to the end date. The following examples illustrate acceptable values for a period slot:

  • [2002, 2006]
  • [2002-01, 2002-01]
  • [2002-01-18 10:32:11, 2006-03-25 10:58:43]
  • [2002-01-18 10:32:11, 2006-03]

Contrary, the following examples illustrate unacceptable values for a period slot:

  • [2006-03-25, 2002-01-18] (ending date is before the starting date)
  • [2006, 2006-10-01] (starting date represents the whole of the year 2006, which includes dates that are after the ending date).

Semantically, a period is considered to include both the starting and the ending time point, e.g. the period [2002-01, 2002-02-12] is considered to include the whole of January 2002 and the twelve first dates of February. This is particularly important for the notion of overlapping periods used in the context of non-multiple temporal types (discussed in the next paragraphs). According to the adopted semantics, periods [2002-01, 2002-02-12] and [2002-02-12, 2002-03] do overlap (since they both include February 12, 2002), while periods [2002-01, 2002-02-12] and [2002-02-13, 2002-03] do not overlap.

The special date values unknown and now may be used for the starting or the ending date of a period (or both). In such a case, no check is performed that the starting date should be before the ending date.

The "t-String" Data Type

The "t-String" data type allows for entering string-typed values that retain their evolution through time. For example, consider the case that we want to retain history of the addresses at which some organization is installed; for this purpose we would use a t-String typed slot named "Address". In the instance editing window, the slot would be rendered as standard Protégé list box, and we would use the "Add Value" button to enter the proper values for the organization whereabouts, each one tagged with a period expressing when the organization was installed at the specific location, e.g.:

[2000-01-30, 2004-12-06]Somewhere, Someplace 23, 12345, Neverland
[2004-12-07, 2006-03-08]Elsewhere, Otherplace 42, 34567, Neverland
[2006-04, now]Anywhere, Anyplace 4, 67890, Neverland

Note that it is allowed to use mixed granularities across periods timestamping different values (i.e. the first two rows use day-level granularity, while the third one employs month-level); the use of "now" and "unknown" is also permitted.

The semantics for the "multiple" check box in the slot property window are modified for the t-String data type as follows:

  1. if the "multiple" check box is not checked, then it is not allowed for any two list entries to have overlapping timestamps i.e. not multiple means that for any given point in time, at most one value is allowed. If entries with overlapping timestamps are detected in the list, the slot value is considered erroneous and the list is highlighted with a red border, while the tooltip text is set to indicate the offending periods.
  2. if the "multiple" check box is checked, then the timestamp overlap check is not performed, effectively allowing any number of list entries to have overlapping timestamps. This may be used to model cases that multiple values for a single point in time are allowed in the real world, such as organizations with multiple installations (e.g. headquarters, agencies, warehouses etc), lists of professors serving in a University department and so forth.

Under these definitions, the value list depicted in the following table is not valid for a t-String slot characterized as "non-multiple" because the second and third rows have overlapping timestamps; the same value list is valid for a t-String slot characterized as "multiple".

[2000-01-30, 2004-12-06]Somewhere, Someplace 23, 12345, Neverland
[2004-12-07, 2006-03-08]Elsewhere, Otherplace 42, 34567, Neverland
[2005-11, 2006-04-18 10:11]Wherever, Everyplace 88, 98765, Neverland
[2006-04, now]Anywhere, Anyplace 4, 67890, Neverland

No overlap check is performed for timestamps using the "now" and "unknown" date representations at either end.

The "t-Boolean", "t-Instance", "t-Integer" and "t-Float" data types

These data types are similar regarding the semantics and operation with the t-String data type, differing only in the type of the data that may be entered. Booleans, in particular, can be entered as "true" or "false", while instances can be picked from the standard Protégé "Select Instance" window.

Notes and known issues

  1. Ontologies using any of the extensions can be opened only by installations that use the t-protégé enhancements. Thus, if you send such an ontology to another user or transfer it to another computer, make sure that the extended software is also available, or it will not be possible to load the ontology in Protégé.
  2. The window allowing for creating, viewing and modifying timestamped instances does not use the standard Protégé instance handling widget.
  3. The current support for the “now” date is quite limited; better support for displaying these dates and for performing validation checks should be provided.
  4. February 29 is accepted for any year, not just leap ones.
  5. Source code implementing the extensions is in need of some cleanup, documentation and beautification.

Please report bugs and suggestions to tprotege@uop.gr

Publications

Publications related to t-Protégé

Visualizing a Temporally-Enhanced Ontology

Katifori Akrivi, Vassilakis Costas, Lepouras Georgios, Daradimos Ilias, Halatsis Constantin
Proceedings of the ACM Advanced Visual Interfaces 06 Conference.

Abstract:
Most ontology development methodologies and tools for ontology management deal with ontology snapshots, i.e. they model and manage only the most recent version of ontologies, which is inadequate for contexts where the history of the ontology is of interest, such as historical archives. This work presents a modeling for entity and relationship timelines in the Protégé tool, complemented with a visualization plug-in, which enables users to examine entity evolution along the timeline.

A Comparative Study of Four Ontology Visualization Techniques in Protégé: Experiment Setup and Preliminary Results

Akrivi Katifori , Elena Torou, Constantin Halatsis, Lepouras Georgios and Costas Vassilakis
Proceedings of the IV 06 Conference.

Abstract:
The continuing need for more effective information retrieval has lead to the creation of the notions of the semantic web and personalized information management, areas of study that very often employ ontologies to represent the semantic context of a domain. Consequently, the need for effective ontology visualization for design, management and browsing has arisen. There are several ontology visualizations available through the existing ontology management tools, but not as many evaluations to determine their advantages and disadvantages and their suitability for various ontologies and user groups. This work presents the preliminary results of an evaluation of four visualization methods in Protégé.

t-Protégé – A Temporal Extension for Protégé

Vassilakis, Costas, Lepouras, George, Katifori, Akrivi, Technical Report TR-SSDBL-06-001, June 2006, Tripoli, Greece

Abstract:
This work presents t-Protégé, a temporal extension to Protégé allowing for creating, managing and visualizing ontologies for time-varying domains.

Citation

This work should be cited as:
Vassilakis, Costas, Lepouras, George, Katifori, Akrivi (2006). t-Protégé – A Temporal Extension for Protégé, Technical Report TR-SSDBL-06-001, June 2006, Tripoli, Greece

Downloads

Downloads for the t-Protégé project:

  • protege.zip is the compiled software for managing temporally enhanced ontologies
  • protege_src.zip is the full souce code tree (includes Parser.jj under directory bin); contains the original Protege source code as well as modifications and additions
  • documentation.doc is the project documentation in Microsoft Word for Windows format
  • documentation.pdf is the project documentation in PDF format
AttachmentSize
Package icon protege.zip1.23 MB
Package icon protege_src.zip1.28 MB
Microsoft Office document icon documentation.doc93 KB
PDF icon documentation.pdf179.26 KB