HuTO Vocabulary Specification

Abstract

HuTO (Human Time Ontology) is an ontology formalized in RDFS allowing temporal annotation using temporal expressions of common speech. This ontology also defines the temporal anchors related to the context and capture temporal changes associated with annotated resources. This makes it possible to query Temporal information from the knowledge base using SPARQL queries.

We distinguish two research domains about handling temporal information in Semantic Web: temporal modeling and temporal data annotation (time scope). Temporal modeling allows to model a temporal expression as a date, a convex interval, a non-convex interval, a relative or absolute temporal expression, etc. For instance, Monday April 18, 2013; every Wednesday; every even month of year 2014. Temporal data annotation allows to represent temporal notions in order to annotate knowledge (expressed as RDF triples) and that preserving the evolution of data (change in value) over time.

Advantages of using HuTO's modeling are:

Table of Contents

1. HuTO in nutshell
2. Examples
3. Classes
4. Object Properties
5. Data Properties
6. Namespace Declarations
7. References

1 HuTO in nutshell

The figure below shows the core classes and properties of HuTO.

1.1 Date, Calendar Date and Unit

In HuTO, main concepts for datation are Datation and TemporalUnit (Fig. 1) which are used to model date and duration. The Datation concept is an abstract concept (which has no instances) from which are derived Date and Duration concepts. Date concept allows to model dates such as the xsd:dateTime (examples a et b). It represents a specific date and time. A time zone expression may be added with TimeZone concept by given the difference with UTC (Coordinated Universal Time). If no time zone value is present, it is assumed to be UTC. Duration concept allows to define duration such as the xsd:duration. It represents a duration of time expressed as a number of years, months, days, hours, minutes, and seconds (example d).

HuTO allows to model deictic dates with GenericDate concept (Fig. 1) which has as sub-concepts GenericDay (Today, etc.), GenericWeek (lastWeek, etc.), GenericMonth (NextMonth, etc.) and GenericYear (ThisYear, etc.).

As temporal units, we consider LunarUnit and SolarUnit. This approach allows to model different calendar units such as Chinese or Hebrew calendars but also Senegalese sociocultural events which all use solar and lunar units. It is possible to define other temporal units such as Maya or Aztec units which will be specializations of TemporalUnit concept. As we are interested in modeling Senegalese sociocultural events, in Figure 1 we detail Gregorian units for which defined granularities range from Century to Second and lunar unit which is used to model religious events in Senegal.

1.2 Instant, Interval and Duration

A temporal element can be an instant, an interval or a duration. We have chosen to represent all temporal elements as intervals modeled using the concept TemporalExp (Fig. 1). Therefore, if the end or the duration of the interval is not specified then the considered interval is the date unit. For instance, the date Friday, August 15, 2014 is considered as 24 hours interval. To specify the beginning and/or the end of an interval, one must use the TemporalExp concept with hasBegin and/or hasEnd properties. To model an interval with duration, TemporalExp concept is also used with hasBegin property to specify the beginning and hasDuration property to specify the duration.

The Cycle concept is used to model non-convex intervals (repetitive). Non-convex intervals are characterized by two entities: the repetition frequency and the convex interval to repeat. This frequency is a sub-concept TemporalUnit which represents the time unit to which the cycle is repeated. The convex interval is connected to the Cycle concept by the relation exp.

Overall, the exp (Fig. 1) property is used to connect a non-convex interval (Cycle) to a convex interval (TemporalExp) or another non-convex interval. It is used also to connect a convex interval to a non-convex interval.

In HuTO, at this moment, we included before and after properties (Fig. 1). They allow us to support relative dates which allows to determine the relative order of events without necessarily determining their absolute date as in the expression The Russian Revolution was after the French Revolution (Example g). Note that it allows us to have two implicit information: 1) the date of the referenced resource and 2) the two Allen relations between resources. In HuTO, before and after properties may be expressed between intervals, between resources and between a resource and an interval.

1.3 Temporal Data Annotation

Temporal data annotation consists in linking data (a resource, a triple or a named graph) to their temporal dimension. Thanks to our modeling choices, two dimensions co-exist: temporal and non-temporal. The temporal dimension is specified using HuTO concepts and the non temporal one, consists in the existing triples in the knowledge base. Thus, information retrieval is facilitated by considering or not the temporal aspect.

Overall, temporal annotation can be associated with a resource, a triple or a named graph. To annotate these objects, we create TemporalAnnotation (Fig. 1) which has two properties. The first one, hasTemporalExp property, allows to represent the temporal expression that annotates the resource. The second one is associated with the resources to annotate as follows:

Note that HuTO also allows to model qualitative temporal notions (Fig. 1). HuTO allows a resource to be used as a temporal reference through the TemporalAnnotation concept. In this case, it is considered as a temporal object which can be used as a time anchor. Thereby, instead of to refer to the occurrence date one uses the annotated resource such as in the example g. It also is possible to model several temporal notions relative to one culture or to the geographical position.

For instance, the notion of PartOfDay is shared across the world but the duration depends to the geographical position. Thereby, the notion of Night means the same thing in Dakar (Senegalese capital) and Sydney but the intervals are not the same. On July 29, 2015 the Night is from midnight to 5am 30 and from 9pm to 11pm 59 at Dakar while in Sydney it is from midnight to 5am 30 and from 6pm 30 to 11pm 59. Thus to model this difference, DakarNight and SydneyNight concepts may be created as sub-concepts of Night concept and each of them will have its own schedule.

Thereby to model temporal domain notions, HuTO's qualitative notions can be used by specifying the good interval relative to geographical position. After that, the notion can be used such as in the example g as a time anchor. The same approach is done relative to the Season. African countries like in Senegal have a tropical season and the duration and period is relative to the geographical position.

HuTO allows to use a resource as a time reference through the concept TemporalAnnotation. So once dated, a resource can be used as a time marker (Example g).

For Temporal data annotation, HuTO is used to annotate: 1) a resource (Example g), 2) a triple (Example e) and 3) a set of triples (Example f)

Object Properties

• after
• before
• every
• exp
• graph
• hasBegin
• hasCentury
• hasDate
• hasDay
• hasDuration
• hasEnd
• hasUTCOffset
• hasHour
• hasMinute
• hasMonth
• hasSecond
• hasTemporalExp
• hasTemporalPosition
• hasTemporalUnit
• hasTimeZone
• hasWeek
• hasYear
• included
• triple
• uri

Data Properties

• even
• lastWeek
• leapYear
• number
• numberOfDay
• position
• positive
• sample
• value

6 Namespace Declarations back to ToC

huto namespace

http://ns.inria.fr/huto/

rdf

http://www.w3.org/1999/02/22-rdf-syntax-ns#

rdfs

http://www.w3.org/2000/01/rdf-schema#

xsd

http://www.w3.org/2001/XMLSchema#

7 References back to ToC

[1] Allen, J. F. 1984. Towards a general theory of action and time. Artificial Intelligence 23, pp. 123-154.

[2] Allen, J. F. and Ferguson, G. 1997. Actions and events in interval temporal logic. In Spatial and Temporal Reasoning. O. Stock, ed., Kluwer, Dordrecht, Netherlands, 205-245.

[3] Pan, F and Hobbs, J. R. 2005. Temporal Aggregates in OWL-Time. In Proceedings of the 18th International Florida Artificial Intelligence Research Society Conference (FLAIRS), Clearwater Beach, Florida, pp. 560-565, AAAI Press.

[4] W3C Working Draft 27 September 2006. Time Ontology in OWL. Available at http://www.w3.org/TR/owl-time

[5] LADKIN P. B. (1987). The Logic of Time Representation. PhD Thesis at the University of California at Berkeley, November.

[6] GUTIERREZ C., HURTADO C., & VAISMAN A. (2005). Temporal RDF. In European Conference on the Semantic Web (ECSW’05) (Best paper award), pp 93–107.

[7] TAO C., WEI-QI WEI , SOLBRIGH H., SAVOVA G. & CHUTE C. G (2010). CNTRO: A Semantic Web Ontology for Temporal Relation Inferencing in Clinical Narratives. In AMIA Annual Symposium Proceedings: 787–791. Available at http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3041418/

[8] Welty C. & Fikes R. (2006). A Reusable Ontology for Fluents in OWL. In Frontiers in Artificial Intelligence and Applications, 150:226–236.

This visual layout has been adapted from the LODE