An Ontology Design Pattern for representing Recurrent Situations
Valentina Anita Carriero, Aldo Gangemi, Andrea Giovanni Nuzzolese, Valentina Presutti
AA N O NTOLOGY D ESIGN P ATTERN FOR REPRESENTING R ECURRENT S ITUATIONS
A P
REPRINT
Valentina Anita Carriero
Department of Computer Science and EngineeringUniversity of BolognaMura Anteo Zamboni 7, 40126 Bologna, Italy [email protected]
Aldo Gangemi
Digital Humanities Advanced Research CentreDepartment of Classical Philology and Italian StudiesUniversity of BolognaVia Zamboni 32, 40126 Bologna, Italy [email protected]
Andrea Giovanni Nuzzolese
Semantic Technologies LaboratoryInstitute of Cognitive Sciences and TechnologiesItalian National Research CouncilVia San Martino della Battaglia 44, 00185 Rome, Italy [email protected]
Valentina Presutti
Department of Modern Languages, Literatures, and CultureUniversity of BolognaVia Cartoleria 5, 40124 Bologna, Italy [email protected]
January 5, 2021 A BSTRACT
In this paper, we present an Ontology Design Pattern for representing situations that recur at regularperiods and share some invariant factors, which unify them conceptually: we refer to this set ofrecurring situations as recurrent situation series . The proposed pattern appears to be foundational,since it can be generalised for modelling the top-level domain-independent concept of recurrence,which is strictly associated with invariance. The pattern reuses other foundational patterns such asCollection, Description and Situation, Classification, Sequence. Indeed, a recurrent situation series isformalised as both a collection of situations occurring regularly over time and unified according tosome properties that are common to all the members, and a situation itself, which provides a relationalcontext to its members that satisfy a reference description. Besides including some exemplifyinginstances of this pattern, we show how it has been implemented and specialised to model recurrentcultural events and ceremonies in ArCo, the Knowledge Graph of Italian cultural heritage.
In this paper, we investigate the concept of recurrence as applied to a series of situations, which recur , i.e. they occurperiodically or repeatedly, either by design (e.g. a festival), or by featuring similar attributes (e.g. a bird migrationpattern); hence the name recurrent situations . Recurrence is associated with invariance . We typically recognisesomething as a known entity when it is invariant under appropriate transformations (in time, space, context, etc.). Thisis a basic biological competence, typically starting in human babies at two months of age. As we recognise an invariantobject, we can also recognise an invariant situation, when multiple sets of related objects feature similar patterns. Whilethis ability in humans develops into foundational concepts such as permanence , similarity and difference , type , etc., and a r X i v : . [ c s . A I] J a n PREPRINT - J
ANUARY
5, 2021is well investigated in cognitive science (see e.g. [1] for links between current cognitive science theories and knowledgerepresentation), recurrence is less studied in ontology design. In particular, recurrence under time transformation is notfully addressed and captured by state-of-the-art patterns and ontologies.In this paper, we deal with constructed and natural recurrence of situations under time transformation. By constructed we mean a recurrence created by social norms (e.g. elections or festivals) or emerging and established in local or evenindividual behavioral patterns (e.g. social meetings or daily routines). By natural we mean a recurrence observed innature (e.g. the raising of the sun, species migrations). By situation we mean any eventuality or structure that someoneintends to talk about (consider, refer to, mention), e.g. a conference, a system configuration, a trial.Existing ontologies mostly model recurrent situations as a special type of event, which represents its (repeatable)occurrences as generic parts of a more general event, neglecting their belonging to a series , intended as a collection,whose members are temporally ordered. This is relevant and characterises the ontological nature of such a unifyingentity.The Recurrent Situation Series Ontology Design Pattern (ODP), explained in the rest of this paper, addresses this issue.This pattern models collections of situations, in which the member situations: (i) are ordered in a temporal sequence,(ii) are separated by time intervals that obey some function (that is in principle computable), and (iii) share at leastone invariance, unifying factor (e.g. a name, a topic, a purpose, etc.). Unifying factors make a recurrent situationseries distinct from others. This is a foundational pattern, since it models the top-level, domain-independent concept ofrecurrence, as applied to a series of situations.The paper is organised as follows: Section 2 introduces the problem addressed; Section 3 describes the pattern at anabstract level; Section 4 presents two actual implementations of the pattern; Section 5 provides two usage examples.Finally, Section 6 reports the state of the art and Section 7 concludes the paper envisaging future work.
The
Palio di Siena is a horse race held twice each year on predefined dates in Siena, Italy. Preceded by other events,e.g. an historical costume parade, the race itself involves ten horses and riders and consists of three laps of
Piazza delCampo . The first horse that crosses the finish line, even without its rider, is declared the winner and awarded a bannerof painted silk ( palio ). Indeed, by riding bareback, more than one rider is usually thrown off his horse each year duringthe turns.The Palio held on 2 July 2019 and the fall of the jockey
Brigante from his horse during that Palio are both situations .Here, we refer to the notion of situation as a generalisation over any relational concept, e.g. events, states, actions. Bothsituations show a kind of repetition: indeed, every year the Sienese horse race occurs, and, most likely, the same can besaid for the riders thrown off their horses. Nevertheless, it is possible, at both common-sense and philosophical level,associate only the first situation with a notion of recurrence that is strictly related to invariance: in recurrent situations,we can recognise a pattern in the iteration, at periodic intervals, of some properties that occur in each situation. Whilethe Palio is explicitly designed, based on a plan, to iteratively occur at specific dates and following regulations (e.g.ten participants, crossbred horses), the falls of the jockeys during the horse race have a high probability, which canbe empirically observed, to occur frequently, possibly at each race, but this observation is statistical in nature and isbased on neither a plan nor a rule that regulates and assures its repetition with specific characteristics and at regulartime intervals. Therefore, the annual race is designed as part of a series with an inherent continuity, while the fall of aspecific jockey is perceived as an episode.There are two main distinctions that can be made between different types of recurrent situations, based on the origin ofthe regular recurrence: (i) natural vs social, (ii) unplanned vs planned.Natural recurrent situations are not constructed by social norms or practices: humans can observe them in very differentways, but their occurrence is not caused by the mere acceptance of social rules. For example, morning sunrise, eveningsunset, heartbeat, periodic migration of groups of animals from a region to another for feeding or breeding, changingseasons during the calendar year. We can be aware or not, we can provide alternative theories, but those things are nothappening because we decree they should exist.Social situations instead are man-made situations that have a specific purpose in a social setting, whose recurrence is setby humans: a world day, a train schedule, festivals, religious holidays, awards, sporting events, parliamentary elections,etc.Planned situations are those defined by a clear plan, which regulates how the things involved in that plan shall be carriedout; on the contrary, unplanned situations have not a predefined setting, and their properties are typically observed aftersome occurrences. 2
PREPRINT - J
ANUARY
5, 2021When we informally (in conversations) refer to a regular repetition of a situation (e.g. an edition of the Palio), the term recurrent event is often used, as we are talking about an event with invariant properties that occurs repeatedly. The term event however may cause confusion as it is attributed in literature several different and very specific meanings. Theconcept situation generalises over them, addressing the large spectrum of possible entities that can recur, either forconstruction or by nature.Going back to our example, the
Palio di Siena is a series of consecutive situations that belong, as members, to acollection characterised by some unifying factors, i.e. properties that give those situations a unity.
In this section, we describe the abstract model of recurrent situation series, which is represented as an Ontology DesignPattern, according to OWL constructs.
Recurrent situation series.
A recurrent situation series is a set of situations repeating regularly over time. It isintended as a collection , since it can be seen as a social object that depends on the collected items, i.e. its members [2].A recurrent situation series is a particular type of collection where all the members are situations: entities flowing intime, either in the physical or social world, located in place, and involving some objects. The members of a recurrentsituation series, i.e. the recurring situations, must all play at least the role of ‘being a member of a recurrent situationseries’, and share one or more common properties.At the same time, a recurrent situation series is a situation itself, intended as a context in which a set of entities iscontextualized based on a description or “frame”. A recurrent situation series is similar to a plan, which defines how thethings involved in that plan (i.e. the specific situations) shall be carried out, or follow some not completely predefinedrules, e.g. where the situations are located, in which time of the year, etc.
Unifying criteria.
The individual (re)occurring situations, while preserving their identity, are kept together and forma new entity, i.e. a recurrent situation series. In [2], collections of items are conceptually unified by unity criteria . Aunity criterion is a concept, a social object created for collecting existing entities, representing a role that can be playedby different entities. Therefore, a recurrent situation series is a collection, whose members are unified by at least oneunifying factor, i.e. organized according to one (or more if it evolves through time) recognizable pattern shared by allmembers. For example, unifying factors of Palio di Siena may be its name and the holding of a horse race in Piazza delCampo. A set of situations can be perceived as a unitary collection when at least one element or property occurs in eachsituation member of the series, making all the situations ascribable to a homogeneous collection. For instance, all thesituations member of a recurrent situation series can be placed in the same location, be about the same topic, be inspiredby the same theme, establish certain activities, involve the same community, etc. A recurrent situation series must haveat least one unifying factor. Considering that, in theory, according to the Description and Situation ODP [3], everysituation satisfies some description. We can expect that all members of a recurrent situation series satisfy the samedescription. Therefore, we can assert that it exists a description that plays the role of unifying factor for a recurrentsituation series. This description is the one satisfied by all situation members of the series.A description is a socially-created object that provides a conceptualisation or a view on a situation. A recurrent situationdescription (RSD hereafter) either defines specific concepts, e.g. repetition interval , or (re)uses existing ones,e.g. location , which are supposed to classify the entities involved in the recurrent situation, e.g. for repetition interval or Siena for location .A RSD looks like a plan for recurrent situations: it unifies the recurrent situations series, describes its member situations,and is satisfied by these situations: this implies that at least some concepts in the RSD classify the entities (e.g. , Siena ) involved in the situations.These unifying factors may be more or less flexible in unifying the recurrent situation series: properties that usuallyoccur in all the situations member of the same collection can undergo change under particular conditions. For instance,yearly recurring situations may be usually held on a specific month of the year, but they still remain members ofthe same series even if one member situation takes place in another period of the year due e.g. to unusual climaticconditions. E.g., the FIFA world cup has always been in May/June/July since 1930, but in Qatar 2021 edition it will beheld in December, since the summer is too hot in Qatar. Moreover, a collection of situations recurring on an annualbasis remains the same entity even if one specific edition is cancelled due to exceptional events.
Temporal sequence.
As opposed to other types of collections, a partly predictable temporal interval holds betweenthe members of a recurrent situation series. The members of the collection are all consecutive situations, thus creating atemporally-ordered sequence, so that each situation may have either previous situations, or next situations, or both,3
PREPRINT - J
ANUARY
5, 2021Table 1: Competency questions answered by the Recurrent Situation Series ODP.ID Competency questionCQ1 Which are the situations of a recurrent situation series?CQ2 Which is the time period elapsing between two situations of a recurrent situation series?CQ3 When is the next situation of a recurrent situation series scheduled?CQ4 What are the unifying criteria of a recurrent situation series?CQ5 Which is the temporal validity of a unifying factor of a recurrent situation series?CQ6 Which is the description satisfied by all the situations of a recurrent situation series?CQ7 Which is the (immediate) next situation in a recurrent situation series?CQ8 Which is the (immediate) previous situation in a recurrent situation series?unless the recurrent situation series, planned to be a series of at least two situations, is interrupted after the first situation,or is never instantiated.In the context of this pattern, the temporal interval is relevant only between members of the same recurrent situationseries, while the sequence between a member situation and any other situation is out of the scope of this pattern.
Regular time period.
A collection of situations can be defined recurrent if its members occurrence are separated byregular time periods. As for natural situations, this time period is observed in nature, while social situations usuallydefine the recurrence in a plan, explicitly. The time period concept works as a parameter that classifies typical setsof values: for example, a “yearly” time period classifies approximated time intervals, e.g. between 11 and 13 monthsinstead of exactly 12.Even considering approximation, there is a difference between the estimated time period, i.e. the time interval thatis planned (for social and planned situation series) or foreseen (for natural and unplanned situation series) when therecurrent situation series starts, and the actual time period, i.e. the time interval that can be observed after the serieshas been instantiated and has multiple members. For instance, to a planned time period of “approximately 2 years”corresponds a measured time interval that represents the average of time intervals actually occurring between thesituations, and can vary significantly if e.g. some yearly situations are postponed or canceled. This measurement couldbe especially relevant to recurrent situations series, when the time period is a requirement to satisfy (e.g. a periodicaljournal), or when a time period different from the expected is due to negative external causes (e.g. natural climatechanges).
The abstract pattern, as described in Section 3, can be formalised and possibly implemented as an Ontology DesignPattern (ODP), in order to be reused in different contexts. In this section, we report as an example two implementationsof this pattern: the ODP that has been published on the
ODP portal [4], and an implementation specific to the CulturalHeritage (CH) domain, in the context of the ArCo ontology network [5]. The Recurrent Situation Series pattern reuses five existing Ontology Design Patterns (ODPs): Collection , Situation , Description and Situation , Classification , Sequence . These patterns have been extracted from DOLCE+DnS Ultra-lite foundational ontology, a lighter version of DOLCE+DnS [6], thus our ODP is also aligned to it. Moreover, allreused ODPs are annotated with the OPLa ontology [7], which facilitates future reuse of this pattern by supporting theprocess to understand and explore the ODP. For instance, OPLa allows us to express that a pattern in a local ontology isa specialisation or a generalisation of another pattern. http://ontologydesignpatterns.org https://w3id.org/arco/ontology/arco rss: http://ontologydesignpatterns.org/wiki/Submissions:Collection http://ontologydesignpatterns.org/wiki/Submissions:Situation http://ontologydesignpatterns.org/wiki/Submissions:Classification http://ontologydesignpatterns.org/wiki/Submissions:Sequence dul: PREPRINT - J
ANUARY
5, 2021The ODP, as submitted to the ODP portal, can be found at http://ontologydesignpatterns.org/wiki/Submissions:RecurrentSituationSeries . A diagram depicting the pattern is shown in Figure 1, and the Compe-tency Questions (CQs) this pattern can answer are reported in Table 1. The general pattern provides a reusable templatefor modelling recurrent situation series. For the concepts of situation and time period, which are not native to thispattern, it reuses the semantics associated with: d0:Eventuality , from the foundational ontology DOLCE-Zero , asa top-level concept for any situation, event or activity, and the formalisation of time periods as defined by the TimePeriod
ODP . In the following, we describe classes and properties of the pattern.Figure 1: Pattern for Recurrent Situation Series, in Graffoo notation (https://essepuntato.it/graffoo/specification/). Arecurrent situation series is related to: its member situations, which are temporally ordered in sequence; one or moreunifying factors; one or more unifying situations, which involve unifying factors and their temporal validity; theestimated and measured time period between two consecutive situations. At least one unifying factor of the series is adescription that is satisfied by each situation member of the series. Moreover, the series satisfies a description definingthe concepts characterising recurrent situation series. RecurrentSituationSeries.
Recurrent situations are represented as individuals of the class :RecurrentSituationSeries , which is modeled as the intersection of the classes dul:Collection and :Situation (axiom 1): a recurrent situation series is both a collection of situations, and a situation on its turn,extending along the time in which its members occur (see below).
RecurrentSituationSeries (cid:118) dul:Collection (cid:117)
Situation (1) hasMemberSituation.
The membership for a collection of recurring situations (a :RecurrentSituationSeries )is restricted to situations only (axiom 2), and a specific object property :hasMemberSituation is provided as apattern vocabulary item. Its inverse property is :isSituationMemberOf . A recurrent situation series may also have 0members, in order to envision series that are never instantiated (axiom 3). As aforementioned, the concept of situation isnot native to this pattern, thus we introduce a generic class :Situation as a subclass of d0:Eventuality (axiom 4). d0: http://ontologydesignpatterns.org/wiki/Submissions:TimePeriod PREPRINT - J
ANUARY
5, 2021
RecurrentSituationSeries (cid:118) ∀ hasMemberSituation.Situation (2)
RecurrentSituationSeries (cid:118) (cid:62) hasMemberSituation.Situation (3) Situation (cid:118) d0:Eventuality (4)
UnifyingFactor. :UnifyingFactor is modelled as a subclass of dul:Concept , since it is a social object forcollecting situations. The :RecurrentSituationSeries is related to the invariant unifying factors by means ofthe object property :hasUnifyingFactor . An existential axiom (axiom 5) states that it exists a dul:Description playing as unifying factor for a recurrent situation series. This description defines concepts ( dul:defines ) thatconceptualise all the situations that are member of the series, thus it dul:isSatisfiedBy them. Therefore, it works asa unifying criterion for the whole series. Moreover, the recurrent situation series itself satisfies another description,which defines all the concepts conceptualising a recurrent situation series (e.g. the general concept of unifying factor,the concept of time period, etc.).
RecurrentSituationSeries (cid:118) ∃ hasUnifyingFactor.dul:Description (5)
UnifyingSituation, isValidIn.
The validity of a unifying factor can be limited to a specific time interval, for criteriathat unify the series for a limited period of time, due to their flexibility or external causes. For modelling the temporalvalidity of a unifying factor, we need a temporally-indexed situation ( :UnifyingSituation ) involving a unifyingfactor (axiom 6). The relation between this unifying situation and the time interval of its validity is expressed throughthe :isValidIn object property (axiom 7).
UnifyingSituation (cid:118) ∃ involvesUnifyingFactor.owl:Thing (6)
UnifyingSituation (cid:118) ∃ isValidIn.dul:TimeInterval (7) hasNextSituation, hasPreviousSituation.
It is possible to represent which is the previous and the next situation of aparticular member of the series by means of the object properties :hasPreviousSituation and :hasNextSituation .The two properties are further specialised into :hasImmediatePreviousSituation and :hasImmediateNextSituation , respectively, in order to associate the situations with their immediate next and previous situations in thesequence.These properties express a sequential relation between situations that are members of the same recurrent situation series,while they do not relate a situation to any other previous or next situation. This intended restriction cannot be expressedin OWL 2 DL [8], because this requires a “diamond” construction with explicit coreferent variables. However, it ispossible to check the local integrity of an instantiation of a :RecurrentSituationSeries by using inference rules.A situation (member of a recurrent situation series) can be linked by the property :hasNextSituation to anothersituation, which is also member of a recurrent situation series, if both situations are member of the same collection.For instance, let us consider the following SPARQL query. A triple asserting a local inconsistency between the tworecurrent situation series will be generated, if they are not the same or if there is not an identity relation ( owl:sameAs )between the two recurrent situation series. Similar SPARQL queries can be executed for the related constraints (e.g.situations connected by :hasPreviousSituation object property). These queries can be embedded in the RDFmodel via SPIN or SHACL . https://spinrdf.org/ PREPRINT - J
ANUARY
5, 2021The datatype property :isTheLastSituation , with range xsd:boolean , expresses whether a situation is the last ofthe series or not, while :situationNumber represents the number of the situation within the sequence (e.g. “2” forthe 2nd situation).
TimePeriod.
While the time period is a core element of this pattern, it can be modelled in different ways, dependingon the local requirements. Thus, we introduce a generic class :TimePeriod . Nevertheless, we propose a possiblesolution for implementing the pattern, by reusing the class tp:TimePeriod from the
TimePeriod
ODP in ourusage examples (see Section 5). A tp:TimePeriod is related to a measurement unit and a measurement value. Thecollection of situations is related to the recurrent time period with the object property :hasTimePeriod (axiom8), which is defined as a property chain [:hasMemberSituation ◦ :hasTimePeriodBeforeNextSituation] .The object property :hasTimePeriodBeforeNextSituation relates a situation, member of a recurrent situationseries, to the approximate (e.g. yearly, monthly, etc.) time period that typically elapses before the next situation,and is different from the actual time interval between two specific situations of the series, which can be derivedfrom the time intervals measured between any two members. :hasTimePeriod is indeed further specialised into :hasEstimatedTimePeriod and :hasMeasuredTimePeriod . RecurrentSituationSeries (cid:118) ∃ hasTimePeriod.TimePeriod (8)
TimePeriodMeasurementUnit.
The tp:TimePeriod is related to the unit for measuring it, tp:TimePeriodMeasurementUnit (e.g. year, month, week), by means of the object property tp:hasTimePeriodMeasurementUnit , and to the time value through the datatype property tp:timePeriodValue ,with range xsd:integer . A second implementation of the pattern is given in ArCo (Architecture of Knowledge) , the Italian Cultural Heritage(CH) knowledge graph [5]. We have been working on formally representing recurrent situations while developing thisKG, which consists of an ontology network and facts on Italian cultural properties, based on the General Catalogue(GC) maintained by the Italian Institute of the General Catalogue and Documentation (ICCD-MiBAC). ArCo adoptsan agile and iterative, pattern-based and test-driven ontology development methodology, called eXtreme Design (XD)[9]. Its requirements are provided in the form of user stories, as scenarios and real use cases, by a growing communityof customers, including ICCD. In particular, two modelling issues led us to implement a solution for representingsituations recurring regularly over time.In the GC catalogue records, which describe Italian cultural properties, it is possible to find information about events(title, place, organizer) involving cultural properties, such as exhibitions. While there is no explicit information aboutpossible recurrent cultural events, by analyzing the data, it is clear that there are many cases of collections of recurrentevents (e.g. “third exhibition", “tenth painting award").The second use case concerns a particular type of intangible cultural heritage: ceremonies, customs and celebrationsrelated to the year cycle (e.g. Carnival, Ramadan) or to the season cycle (e.g. popular belief, science, myth, phenomenarelated to specific periods of the year), which recur regularly and whose periodicity is explicitly referred to in the GCcatalogue records (e.g. annual, every two years, three times a year).The Recurrent Situation Series ODP has been indirectly reused (i.e. has been reused as a template) in the CulturalEvent ontology module. It has been specialised in order to represent collections of situations that are cultural events( cis:CulturalEvent ) and exhibitions ( a-ce:Exhibition ). In this section we introduce two use cases, and show how they can be addressed by the Recurrent Situation Series ODP.The first example is a workshop, planned to occur periodically; the second example concerns a collection of recurrentsituations observed in nature: annual animal migrations. tp: https://w3id.org/arco a-ce: https://w3id.org/arco/ontology/cultural-event cis: http://dati.beniculturali.it/cis/ PREPRINT - J
ANUARY
5, 2021
The Workshop on Ontology Design and Patterns (WOP) is a series of workshops that are co-located with ISWC(International Semantic Web Conference). ISWC started in 2002, WOP in 2009. Both the conference and the workshoprecur yearly, between October and November, hence there is a time period of about one year between two situationsof the same series. Nevertheless, it can happen that one edition is skipped: indeed, WOP has not been held in 2011.The location of these events may change every year. Usually, 3 geographical regions are involved: Europe, America,Asia/Oceania are sequentially alternated. However, unforeseen conditions can alter this pattern: in 2020, due toCOVID-19 pandemic, they are organised as virtual events. The Workshop on Ontology Design and Patterns can be seenas a unitary collection on account of many (more or less) invariant factors: it is co-located with ISWC, its main topic ismodular and pattern-based design, it involves a community mostly formed by computer scientists and ontologists, it isorganised by the Ontology Design and Patterns Association (ODPA), etc. Some factors are common to a subset of thesituations member of the series, e.g. the name of the workshop changed 3 times: (i)
Workshop on Ontology Patterns from the first edition to 2012, (ii)
Workshop on Ontology and Semantic Patterns from 2013 to 2016, and (iii)
Workshopon Ontology Design and Patterns since 2017. Moreover, the workshop recurs regularly: its estimated time period is ofabout 1 year, but the measured time period slightly differs, since it was not held in 2011, and its dates are not predefined.Figure 2 depicts how WOP can be modelled by reusing this ODP. In Subfigure 2a, the series collecting the annualworkshops is an instance of :RecurrentSituationSeries , and is related to its members (e.g. WOP 2009, WOP2010 and WOP 2012). The sequential relations between these (immediate) consecutive situations are modelled by hasNextSituation and hasImmediateNextSituation , and their inverses. Since the beginning of the series, threeunifying factors have been valid: the topic of the workshops ( ex:pattern-based-design ), the organising committee( ex:wop-organisation ), and being held as joint event of ISWC ( ex:co-location-iswc ). So far, these unifyingfactors do not necessarily need the specification of a temporal validity. Differently, the name of the workshops can beconsidered a unifying factor, however it has changed over time. This can be modelled by introducing a :UnifyingSituation to specify the time interval in which e.g. the current name of the series is valid ( ex:2017-present ). Theseries is related to its estimated time period of ex:1year .Subfigure 2b depicts the description satisfied by the members of the ex:wop-series , i.e. ex:wop-description .This description defines a number of concepts (the topic, the role of chair) that can be used for classifying the entitiesinvolved in the member situations (each workshop edition). For instance, the concept ex:topic classifies the entity ex:pattern-based-design , which is the main topic of the workshop(s). Indeed, the factors unifying the whole seriesmay be classified by some or all the concepts defined in this description. Finally, the WOP series satisfies a descriptionthat defines the core concepts of recurrent situation series ( ex:recurrent-situation-series-description ), suchas the time period, the role of being a situation member of a series, the unifying factor.
Animal migration is the regular long-distance movement of animals, on an annual or seasonal basis, that may be causedby local climate, the season of the year, and food availability. The Arctic tern (
Sterna paradisaea ) is a long-distancemigrant seabird: it is famous for periodically flying from the northern hemisphere, where it breeds, to the SouthernOcean and back again each year. In this way, it lives two summers a year. It departs the breeding site (Arctic) aroundAugust and starts its flight back from the Antarctic around March. The migration of the Arctic tern is a unitary collectionof situations since its member situations involve all the members of a specific bird species, happen along the North-Southaxis, depend on the season of the year in a specific place, etc.Figure 3, depicts how the migration of the Arctic ster is modelled with this ODP. The ex:arctic-tern-migration is an instance of :RecurrentSituationSeries , and represents the annual migration of the seabird. The seriesincludes as members all the annual migrations, e.g. the ex:arctic-tern-migration-2019 is the member situationrepresenting the migration from North to South and back again happened in 2019. This situation actually consistsin two parts ( dul:hasPart ), which are also situations: the flight from North to South in August 2019 ( ex:arctic-tern-migration-ns-2019 ) and the flight from South to North in March 2019 ( ex:arctic-tern-migration-sn-2019 ). These situations are in turn members of two separate recurrent situation series, also with an annual timeperiod: the collection of all the flights from North to South and the collection of the ones from South to North. The ex:arctic-tern-ns-migration and the ex:arctic-tern-sn-migration are series part of the annual arctic ternmigration series. 8
PREPRINT - J
ANUARY
5, 2021 (a) The Workshop on Ontology Design and Patterns (WOP) is a recurrent situation series and is related to its first three consecutivesituations; three unifying factors and one unifying situation with a limited temporal validity; the estimated time period between twoconsecutive events.(b) The situations member of the WOP series satisfy a description; this description defines the concepts (topic, chair) characterisingeach workshop, and provides a unifying factor to the series. The series, in turn, satisfies a description defining the core concepts(time period, unifying factor, series member) of recurrent situation series.
Figure 2: The Workshop on Ontology Design and Patterns (WOP) is a recurrent situation series.
Most work about recurrence in situations focuses on different types of recurrent situations and of constraints that bounda recurrent situation. [10] and [11] investigate calendar-based periodic situations, e.g. repeating time intervals suchas “the first Friday each month”. [12] distinguishes between strongly periodic events, intermittent events and nearlyperiodic events, with respect to the strength and the regularity of intervals between different occurrences. [13] defines aperiodic pattern as the repetition of a component (e.g. an event) over time, such as daily glucose measurements. Bothlocal (referring to a single component, e.g. the duration of one event of the pattern) and periodic (referring to the set of9
PREPRINT - J
ANUARY
5, 2021Figure 3: The Arctic tern migration is a recurrent situation series, with the annual migrations of the Arctic terns asmembers. Each annual migration consists in two parts, the migration from North to South and the migration from Southto North, which are situations members of two separate recurrent situation series. Finally, these recurrent situationseries are part of the annual Arctic tern migration.repeating events, e.g. the duration of the whole pattern) constraints can be applied to periodic patterns, which can bealso split into sub-patterns applied only to a subset of the repeating events. Constraints related to single members of theperiodic series and to the series as a whole are also considered in [14].The Recurrent Situation Series pattern follows the approach adopted by [14], where repeated events are considered ascollections of events of the same type, and can be infinite in case no duration of the collection has been defined. Here,the notion of recurrent situation is related to the idea of a coherent series of events that become part of a pattern bymeans of their iteration, at regular but not necessarily equal time intervals.The concept of collection is studied in [2], but the focus of the authors is on collections of endurants , i.e. entities thatare directly localised in space such as objects and substances, while they do not deal with the properties peculiar to acollection of situations, as entities directly localised in time ( perdurants ).There exist some examples of ontologies modelling recurrent situations, but in most cases they are modeled as aparticular type of event, rather than as a collection of situations or events.In DBpedia , recurrent situation series such as annual film award ceremonies are represented as events, and DBpediaontology models concepts such as dbo:Olympics , dbo:Tournament and dbo:MusicFestival as subclasses of dbo:Event , with an overlap between the series and its different editions. For instance, the Palio di Siena is an instanceof dbo:SportsEvent , and there are no links to specific resources representing its periodic editions. DBpedia ontologyalso derives classes from YAGO ontology, which in turn are automatically derived from Wikipedia and WordNet, suchas yago:WikicatRecurringEventsEstablishedIn1895 , where the semantics of recurrence is only implied bythe name of the class. DBpedia instances of recurrent situations are usually related to their frequency (e.g. “biennial,every two years”) through the dbp:frequency datatype property. https://dbpedia.org/ dbo: http://dbpedia.org/ontology/ http://dbpedia.org/resource/Palio_di_Siena yago: http://dbpedia.org/class/yago/ dbp: http://dbpedia.org/property/ PREPRINT - J
ANUARY
5, 2021Wikidata defines the concept of recurring event ( wd:Q15275719 ), “event recurring at an interval”, as a specific typeof event with an associated time period ( wdt:P2257 ) e.g. 4 years for the Olympic Games. wd:Q15275719 is furtherspecialised based on its frequency (e.g. biennial event) or its category (e.g. recurring sporting events). Wikidata relatesdifferent editions to their recurring event through the object property part of ( wdt:P361 ). Being this property transitive,a sub-event that is part of a specific edition (e.g. a talk by an author of a scientific paper that is part of an edition of anannual conference), would also be considered part of the annual conference as a recurring situation.A mereological approach is less clear in distinguishing typical features of recurrence, e.g. a recurrent situation mayhave no members at all and still be a recurrent situation, but, in this case, a mereological whole would be empty, i.e.without parts, which is both formally and intuitively impossible. Moreover, the intermittent presence of the situationwould make it a part of a maximal continuous situation, which is not the same as a collection of specifically identifiedsituations. In other words, if we admit that recurrence is a temporal mereological concept, we need also to admitmaximal mereological wholes for any set of situations that have some temporally ordered similarity, which in theconsidered use cases would be an overcommitment.As part of the pending section schema.org defines the class schema:EventSeries : “a collection of events thatshare some unifying characteristic”, modelled as a subclass of schema:Event and schema:Series . Included eventsare linked with the series using the schema:superEvent property, which expresses a part-whole relation, thus, asin Wikidata, both the series and the related events are modelled as events, being the latter part of the former. Evenif the comment of the class cites the theme, location and organisers as typical examples of unifying characteristics,these are not modelled in the ontology. schema:eventSchedule associates a series of repeating events with a schema:Schedule , which is the core class for defining the recurrence of an event series, i.e. its repeating time period. schema:repeatFrequency defines the frequency ( schema:Duration ) at which the events will occur according to aschedule, while schema:repeatCount represents the number of times a recurring event will take place. In this paper, we explained an abstract pattern to represent situations that recur regularly over time, in a temporalsequence, and are members of a unitary collection since they share some invariant properties. This pattern can begeneralised in order to represent the more general concept of recurrence. We also discuss two implementations of thispattern as an Ontology Design Pattern (ODP), a small ontology to be reused in different contexts. This pattern has beencreated for addressing the requirement of modeling cultural events, exhibitions, traditional ceremonies, festivals, etc.However, it seems to be useful for representing all recurring situations, which can be distinguished depending on theorigin of their regular intermittence: (i) not artificial situations whose regular recurrence can be observed in nature byhumans, but has a natural origin (e.g. the sunrise every morning or periodic migrations of animals); (ii) man-madesituations where the regular repetition and the unifying factors are set by humans (e.g. a world day or a train schedule);(iii) situations created by humans with a periodicity regulated by a clear purpose, as part of a workflow (e.g. a medicalprescription defining time intervals to take a medicine or periodical supply of services).In our future work, we plan to study all types of recurrent situations, to thoroughly investigate the general concept ofrecurrence, identifying possible new features to be represented, and to experiment and evaluate other possible patternsfor recurrent situations, alternative to the one presented in this paper.
References [1] A. Gangemi, Closing the Loop between knowledge patterns in cognition and the Semantic Web,
Semantic Web (1) (2020), 139–151. doi:10.3233/SW-190383.[2] E. Bottazzi, C. Catenacci, A. Gangemi and J. Lehmann, From collective intentionality to intentional collectives:An ontological perspective, Cognitive Systems Research (2–3) (2006), 192–208.[3] A. Gangemi and P. Mika, Understanding the Semantic Web through Descriptions and Situations, in: On TheMove to Meaningful Internet Systems 2003: CoopIS, DOA, and ODBASE - OTM Confederated InternationalConferences, CoopIS, DOA, and ODBASE 2003, Catania, Sicily, Italy, November 3-7, 2003 , R. Meersman, wd: wdt: Section with ontology terms still work-in-progress, subject to possibly significant change:https://schema.org/docs/pending.home.html schema: http://schema.org/ PREPRINT - J
ANUARY
5, 2021Z. Tari and D.C. Schmidt, eds, Lecture Notes in Computer Science, Vol. 2888, Springer, 2003, pp. 689–706.doi:10.1007/978-3-540-39964-3_44.[4] V.A. Carriero, A. Gangemi, A.G. Nuzzolese and V. Presutti, An Ontology Design Pattern for representing RecurrentEvents, in:
Proceedings of the 10th Workshop on Ontology Design and Patterns (WOP 2019) co-located with the18th International Semantic Web Conference (ISWC 2019), Auckland, New Zealand , K. Janowicz, A.A. Krisnadhi,M.P. Villalón, K. Hammar and C. Shimizu, eds, CEUR Workshop Proceedings, Vol. 2459, CEUR-WS.org, 2019.[5] V.A. Carriero, A. Gangemi, M.L. Mancinelli, L. Marinucci, A.G. Nuzzolese, V. Presutti and C. Veninata, ArCo:The Italian Cultural Heritage Knowledge Graph, in:
The Semantic Web - ISWC 2019 - 18th International SemanticWeb Conference, Auckland, New Zealand, October 26-30, 2019, Proceedings, Part II , C. Ghidini, O. Hartig,M. Maleshkova, V. Svátek, I.F. Cruz, A. Hogan, J. Song, M. Lefrançois and F. Gandon, eds, Lecture Notes inComputer Science, Vol. 11779, Springer, 2019, pp. 36–52. doi:10.1007/978-3-030-30796-7_3.[6] A. Gangemi, N. Guarino, C. Masolo, A. Oltramari and L. Schneider, Sweetening ontologies with DOLCE, in:
International Conference on Knowledge Engineering and Knowledge Management , Springer, 2002, pp. 166–181.[7] P. Hitzler, A. Gangemi, K. Janowicz, A.A. Krisnadhi and V. Presutti, Towards a Simple but Useful OntologyDesign Pattern Representation Language, in
CEUR Workshop Proceedings , Vol. 2043, CEUR-WS.org, 2017.[8] R. Hoekstra and J. Breuker, Polishing diamonds in OWL 2, in:
International Conference on Knowledge Engineer-ing and Knowledge Management , Springer, 2008, pp. 64–73.[9] E. Blomqvist, K. Hammar and V. Presutti, Engineering Ontologies with Patterns - The eXtreme Design Method-ology, in:
Ontology Engineering with Ontology Design Patterns - Foundations and Applications , P. Hitzler,A. Gangemi, K. Janowicz, A. Krisnadhi and V. Presutti, eds, SSWS, Vol. 25, IOS Press, 2016, pp. 23–50.doi:10.3233/978-1-61499-676-7-23.[10] P.B. Ladkin, Primitives and Units for Time Specification., in:
AAAI , Vol. 86, 1986, pp. 354–359.[11] D. Cukierman and J. Delgrande, Characterizing temporal repetition, in:
Proceedings Third International Workshopon Temporal Representation and Reasoning (TIME’96) , IEEE, 1996, pp. 80–87.[12] A. Tuzhilin and J. Clifford, On Periodicity in Temporal Databases,
Information Systems (8) (1995), 619–639.[13] S. Chakravarty and Y. Shahar, Specification and detection of periodic patterns in clinical data, in: Fourth Workshopon Intelligent Data Analysis in Medicine and Pharmacology (IDAMAP-99) , Citeseer, 1999, pp. 20–31.[14] P. Terenziani, Toward a Unifying Ontology Dealing with Both User-Defined Periodicity and Temporal ConstraintsAbout Repeated Events,
Computational Intelligence18