International Conference on Conceptual Modeling

Advances in Conceptual Modeling pp 3-13 | Cite as

PersonLink: An Ontology Representing Family Relationships for the CAPTAIN MEMO Memory Prosthesis

  • Noura Herradi
  • Fayçal Hamdi
  • Elisabeth Métais
  • Fatma Ghorbel
  • Assia Soukane
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 9382)


In the context of the CAPTAIN MEMO memory prosthesis for elderly, we propose the PersonLink ontology for modeling, storing and reasoning on “family relationships” links. Rules are provided to infer new links and/or check inconsistencies in the inputs. On the one hand PersonLink is as generic as possible and is integrated in the linked data formalisms; on the other hand a prosthesis has to be adaptable to users. Thus the PersonLink ontology defines rigorously and precisely family relationships, and takes into account the differences that may exist between cultures/languages, including new relationships emerging in our societies nowadays. The transition from one culture/language to another one cannot be solved with a simple translation of terms, but refers to a meta-ontology and associated mechanisms.


Vocabulary Family relationships Memory prosthesis Multilingual ontology Linked data 

1 Introduction

In 2050, \(30\,\%\) of people from the European countries will be at least 65 years old. Memory troubles will be one of the major disabilities these people will suffer from. A survey of literature suggests that cognitive decline begins averagely before the age of 60 [16]. The prevalence of diseases involving memory impairment such as Alzheimer Disease is directly correlated with age, increasing to \(50\,\%\) in people older than 90 years; memory loss is very progressive and the first symptoms begin many years before the person becomes dependent.

According to [15], participants with memory impairment who have used a PDA1 had their independence in daily tasks increased, with a benefit reminding even if they stopped using the PDA. Ergonomic aspect study and “Design for All” orientation will allow the use of the assistant by elderlies. It is the episodic memory of the elderly that becomes first failing; while the implicit memory - based on routines and automatisms - remains longer preserved, that enable the possibility of creating new connections for durably use of a numeric assistant.

In this context, we are currently developing the CAPTAIN MEMO memory prosthesis that acts as a memory-aid application for elderly people. This prosthesis stores personal data and can be connected to the pervasive environment, using various sensors and localisation systems in order to supply a set of services indoor and outdoor. Among these services, one is devoted to “remember things about people”, i.e. retrieving a person by navigation in the family/entourage tree, retrieving a person according to criteria, retrieving a person facing the camera, retrieving information about a person (name, family or conviviality relationship, age, preferences, gifts exchanged, favourite meals, recent events, shared events, etc.).

The CAPTAIN MEMO memory prosthesis aims at taking into consideration incomplete and inconsistent data. Rules are used either for checking or for deducing. Based on ontologies and semantic web languages, it can (i) store and semantically organize the information; (ii) deduces new facts from the given ones; (iii) check inconsistencies in the input data due to mnesic discordance. In order to store the family and entourage of the user, the modeling is endorsed to an ontology of family and convivial links, which constitutes the backbone of personal information storage.

The works presented in this paper consist in the definition of the PersonLink new ontology for representing and reasoning on the family relationships, that is adaptable to the culture of the person and aware of new concepts of this area. PersonLink gives a precise description of family relationships (e.g., parent/child, grandparents, uncles, cousins, etc.).

The paper is organized as follows. In the next section, we present motivating examples. In Sect. 3 we present some related works and in Sect. 4 the PersonLink ontology. In Sects. 5 and 6 we present experiments on the Captain Memo prosthesis and others performed in the context of linked data. Finally, we conclude and give some perspectives in Sect. 7.

2 Motivating Examples

The memory prothesis has to be adaptable to its users. However family links are completely dependent on the culture and the language [1].

For instance, the surrogate mother carries the child of a couple who gave its embryos. In the United States, there is no federal legislation for Surrogacy. Each state has its own rules, based on jurisprudence. So the Surrogacy is licensed in 14 states. The term used in English for this mother is “Surrogate”. In France, the law no 94–653 of 29 July 1994 prohibits surrogacy. However there exists a term in French which is “Mère porteuse”.

Some concepts may not exists in certain languages, for example “Godmother” does not exist in some cultures/languages. Even if concepts are similar in different languages/cultures, they may differ in their constraints. For instance, depending on the culture, the “spouse” relationship may be defined between 1 man and 1 woman, or between 1 man and several women, etc.

Thus, it is not possible to carry out a translation of terms to switch from a language/culture to another one, since the concept could not exist or could have another definition in the target culture.

3 Related Works

Different ontologies have been proposed to describe family relationships in the web. The most famous one is FOAF [4]. It defines relationships through the predicate “foaf:knows” that links together two individuals; we can see the UML2 Class Diagram on Fig. 1 3. However, this representation is very limited because it does not provide the nature (e.g., family, friendship, etc.) of these links.

The Relationship4 ontology [5] which extended FOAF, introduced several sub-properties to the property “foaf:knows”, that provide some terms representing parenthood, childhood, siblinghood and a generic term representing marriage “SpouseOf”. The Agrelon5 (Agent Relationship Ontology) ontology [9], presents a more precise set of terms that distinguish between the different types of relationships. For example, siblings and half siblings can be distinguished by the two distinct properties “hasSibling” and “hasHalfSibling”. The Relationship and Agrelon ontologies bring more clarity to the relationships. Nevertheless, they remain very generic, lack precision, and they do not support multiculturalism. In [10], Yutaka Matsuo et al. describe the human relationships by considering that every relationship between persons is either belonging to one or more specific events or sharing common properties. In [13], the relationships depend on the context. The authors propose an ontology based on the D & S (Descriptions and Situations) framework. It defines context-specific relationships. This is an interesting solution for making flexible the representation of the relationships, but it cannot permit logic reasoning on interpersonal relationships. The Bio6 ontology, aims to describe biographical information about people.There are some relationships that may be interesting, such as “Father”, “Mother”, etc. However, the Bio ontology is intended to store events (e.g. birth, marriage, etc.) and not links (e.g. father, wife, niece, etc.) since the persons are stored as partners of the same event and do not have a direct link.
Fig. 1.

FOAF (class diagram (UML))

To conclude, the current ontologies offer very short and generic definitions to describe interpersonal relationships. In addition to the lack of precision, the majority of existing ontologies are in English. However, ontologies should be used in different cultures. The PersonLink ontology that we propose deals with these issues by considering the culture/language aspect.

4 The PersonLink Ontology

Our main objective is to represent interpersonal relationships in a precise manner. The Fig. 2 shows an excerpt of the class diagram of the PersonLink ontology with English concepts. However, in the full version there are 12 links between two Person classes, 23 links between the Female class and the Person one, 2 links between two Female classes, 6 links between Female class and Male Class, 22 links between Male class and Person class, 6 links between Male class and Female one and 2 links between two Male classes.

Furthermore the PersonLink ontology represents and defines the concepts according to the considered culture, and expresses them by using terms of the appropriate language. To do this, the first step consists on considering culture in the definition of the concept. For each culture, we look at whether the concept exists or not. If it exists, we describe it using its definition in this culture/language. So, for our ontology, if the concept exists in the culture, a term is assigned to it using the language related to this culture. If the concept does not exist, the term is \(\emptyset \). Note that synonyms might exist in a same language, in this case all the terms are represented. We obtain as a result a kind of sparse ontology that we have called “lace ontology” because it contains many null values as we show in Fig. 3. Then, from this precise definition of concepts related to culture, we proceed to the formal representation of these relationships using the fragments of OWL2 [6] corresponding to the description logic \(\mathcal {SROIQ(D)}\) [7]. Finally, we enrich these relationships by a set of DL-safe rules [14] (ensuring decidability) to have more inference possibilities.
Fig. 2.

Excerpt of PersonLink (class diagram (UML))

4.1 The Lace Meta-ontology

The problem that arises in the representation of properties that describe interpersonal relationships, lies in the description of the property in two different languages/cultures.

Taking the example of the property defining cousins, in French there exist two specific terms that represent this relationship according to the gender of cousin. However only a generic term exists to define this relationship in the English language/culture.

In the PersonLink ontology, we define each concept with a unique number, so each number represents a concept defining a relationship. This will allow us to have a hierarchy with multiple levels of accuracy which combines different languages/cultures. We can move from one concept to another in the level of accuracy (vertically), and therefore from a culture/language to another. Besides, the true meaning of the concept represented by a term for each language (obviously, if it exists in the associated culture) is preserved. We get as a result a lace meta-ontology (because of the null values it may have) of concepts with their representations in different cultures/languages.
Fig. 3.

Excerpt of the meta-ontology for cousinhood in English and French.

The concepts represented in the lace meta-ontology for the cousinhood relationship of a person, shown in Fig. 3, have the following definitions:
  • Concept \(\#2\): the descendant (regardless of gender) of the uncle or of the aunt (both mother’s or father’s side);

  • Concept \(\#2.1\): the female descendant of the uncle or of the aunt (both mother’s or father’s side);

  • Concept \(\#2.2\): the male descendant of the uncle or of the aunt (both mother’s or father’s side).

4.2 Formal Definition of the Relationships

Interpersonal relationships in the PersonLink ontology are represented in a structured way with the OWL2 language that corresponds to the description logic \(\mathcal {SROIQ(D)}\). The use of OWL2 is privileged because it provides a high expressiveness and allows us to represent relationships that we were not able to represent in OWL1. In addition, OWL2, with this logic description, allows semantic reasoners to verify the consistency of data, to derive new knowledge or to extract information already present. Besides, the reasoning in OWL2 is complete and decidable. In predicate logic, the hierarchy represented in Fig. 3 means that:
$$\begin{aligned} 2(?x,?y) \Leftrightarrow 2.1(?x,?y) \vee 2.2(?x,?y) \end{aligned}$$
In order to have more precision about the relationship going from the most generic concept (“cousinOf” in our example) to a specific one (“cousineDe”) (which means: a female cousin), we have to get more information about the instance (female in our example) from the knowledge base of the CAPTAIN MEMO Memory Prosthesis. The specific concept would be inferred from the SWRL [8] (DL-Safe) rules that we would have previously created:
$$\begin{aligned} 2(?x,?y)\wedge Female(?x) \Rightarrow 2.1(?x,?y) \end{aligned}$$
We note that we need the information Female(?x), to deduce that the type of the relationship 2(?x, ?y) (“cousinOf”) is, in this case, the 2.1(?x, ?y) relationship (in French “cousineDe”). We define these relationships in the description logic \(\mathcal {SROIQ(D)}\), with a set of constraints \(\mathcal {K}\). For example, the definition of the cousinhood relationship is the following:
$$\begin{aligned}&\mathcal {K}=\{\exists cousinOf.\top \sqsubseteq Person, Sym(cousinOf), Irr(cousinOf),\qquad \,\,\, \\&\qquad \qquad cousinDe \sqsubseteq cousinOf, Irr(cousinDe), \qquad \qquad \qquad \qquad \qquad \\&cousineDe \sqsubseteq cousinOf, Irr(cousineDe), cousineDe \equiv cousinDe-,\} \end{aligned}$$
We applied the same method to define all the other relationships. The PersonLink ontology is available through a dereferenceable URI (and thus, would be referenced by the Linked Open Vocabulary) at:

5 Validation of PersonLink with Captain Memo

The CAPTAIN MEMO prosthesis [11, 12] has been developed to help elderly to palliate mnesic problems. It acts as a memory-aid application for elderly people. This prosthesis stores personal data and can be connected to the pervasive environment using various sensors and localization systems in order to supply a set of services indoor and outdoor.
Fig. 4.

Graphic editor using PersonLink

Fig. 5.

Inputs through menu

PersonLink is used for modeling, storing and reasoning on “family relationships” links. Figure 4 shows the graphical editor, used for display and search. On Fig. 5 we can see a part of the menu for textual input. The user can also input persons and family relationships through a vocal interface.

In spite of a unified internal representation, a language has to be chosen for the interfaces (we have chosen French for Fig. 4 and English for Fig. 5).

Rules are provided to infer new links and/or check inconsistencies in the inputs. Deducing new facts from ones given by users, and checking inconsistencies in the input data due to mnesic discordance are very important in the case of the CAPTAIN MEMO application targeting elderly persons.

Given the input graph of Fig. 4, the system will for example:
  • Automatically deduce that Jad Cane is the grand-child of François Courtois;

  • Suggest and confirm through a dialog that Marie-Madeleine Courtois could be Jean Courtois’s daughter;

  • Prevent from inputting that Jad is François Courtois’s brother.

6 Validation of the Deduction Mechanism on Large Scale Linked Data

On a larger scale, to test the validity of our reasoning approach, we have taken as a sample persons described in context of the Linked Open Data (LOD). We chose as dataset Freebase, which is a large collaborative knowledge base built mainly from data provided by its community members. We show as relationship example, the cousinship relation presented in Sect. 4. First, we searched in the Freebase ontology, properties that could express the “cousinOf” relationship. In the current versions, Freebase do not uses this property to express that kind of relationship. However, we found other relationships (parental and sibling) that, combined, could be used to express implicitly the cousinship relation. Hence, we extract from Freebase, entities that are linked to each other by a parent and/or sibling relationships, as well as the relationships themselves. This extraction process is done automatically using scripts running MQL7 queries. The obtained results are presented in Table 1:
Table 1.

Entities having parent, child, and sibling relationships in Freebase


Freebase relationship

Number of entities










Sibling’s child




Table 2.

Inferred relationships using Freebase properties

FreeBase relationship

Number of entities

Inferred relationship

Number of entities

Null gender value





















We extracted 2000 entities “person” that have parent relationship. From theses entities, 5155 children and 1815 siblings are generated. There too, no sibling’s child or cousin relationships were found. In the result of this test, we note that the sibling’s children and the person children could be candidates to be cousins. Thus, we integrated entities and relationships that we obtained, on the PersonLink ontology to populate our knowledge base, then we applied automated reasoning to get new relationships. The reasoner is able to infer much more rigorous relationships. For instance, the “fatherOf” and “motherOf” relationships are inferred by exploiting the “parent” and “gender” properties describing Freebase entities. The “daughterOf” and “sonOf” relationships are inferred by exploiting the “children” and “gender” properties. We note that there are some people with null gender values. The results we got are presented in Table 2:

7 Conclusion and Future Works

In this paper we have presented a new ontology called PersonLink, that enables to represent family relationships in the context of a memory prosthesis. PersonLink provides a precise definition for each relationship and takes into consideration the culture/language aspect.

We also presented the notion of lace meta-ontology that facilitates the expression, in multiple cultures/languages of each relationship and allows to switch between languages and find the right terms expressing the relationships. A set of inference rules allow to deduce new links from the given ones, and to check the consistency of inputs, that is particularly useful in the case of a memory prosthesis for users having memory impairments. The current version of PersonLink includes 3 classes (Person, Male and Female), 86 properties, and 582 SWRL rules.

Representing and dealing with family relationships is an important issue on the web; in the context of Linked Data [2, 3], a huge and growing number of data representing persons and relationships between them are published (e.g., data available in DBPedia8, Freebase9, Yago10, etc.). Tests of Personlink conducted on the Freebase dataset show that the use of PersonLink enables inferring much more rigorous relationships than those already present in these datasets. Moreover these tests ensure the scalability of our reasoning mechanism.

Future works will be mainly devoted:
  • To enrich the ontology with convivial links between people (neighbours, friends, care givers, etc.);

  • To enhance deducing rules with context;

  • To take into account time variance.




This research has been partially funded by the “\(Ville \ de \ Paris\)” under the VIVA project (“Vivre à Paris avec Alzheimer en 2030 grâce aux nouvelles technologies”).

The authors wish to thank the reviewers for their evaluable comments.


  1. 1.
    Barry, L.: La parenté. Folio essais (2008)Google Scholar
  2. 2.
    Berners-Lee, T.: Design Issues: Linked Data. 1 March 2012.
  3. 3.
    Bizer, C., Heath, T., Berners-Lee, T.: Linked data - the story so far. Int. J. Semant. Web Inf. Syst. 5(3), 1–22 (2009)CrossRefGoogle Scholar
  4. 4.
    Brickley, D., Miller, L.: FOAF vocabulary specification 0.97. Namespace document, January 2010.
  5. 5.
    Davis, I., Vitiello, E.J.: RELATIONSHIP: a vocabulary for describing relationships between people. Technical report, (2005)Google Scholar
  6. 6.
    Hitzler, P., Krötzsch, M., Parsia, B., Patel-Schneider, P.F., Rudolph, S.: OWL 2 Web ontology language primer (Second Edition). W3C Recommendation, World Wide Web Consortium (December 2012).
  7. 7.
    Horrocks, I., Kutz, O., Sattler, U.: The even more irresistible SROIQ. In: Doherty, P., Mylopoulos, J., Welty, C.A. (eds.) KR, pp. 57–67. AAAI Press (2006)Google Scholar
  8. 8.
    Horrocks, I., Patel-Schneider, P.F., Boley, H., Tabet, S., Grosof, B., Dean, M.: SWRL: a semantic web rule language combining owl and ruleml. W3c member submission, World Wide Web Consortium (2004).
  9. 9.
    Litz, B., Löhden, A., Hannemann, J., Svensson, L.: AgRelOn – an agent relationship ontology. In: Dodero, J.M., Palomo-Duarte, M., Karampiperis, P. (eds.) MTSR 2012. CCIS, vol. 343, pp. 202–213. Springer, Heidelberg (2012) CrossRefGoogle Scholar
  10. 10.
    Matsuo, Y., Hamasaki, M., Mori, J., Takeda, H., Hasida, K.: Ontological consideration on human relationship vocabulary for FOAF. In: Proceedings of the 1st Workshop on Friend of a Friend, Social Networking and Semantic Web (2004)Google Scholar
  11. 11.
    Metais, E., Ghorbel, F., Herradi, N., Hamdi, F., Lammari, N., Nakache, D., Ellouze, N., Gargouri, F., Soukane, A.: Memory Prosthesis. In: Non-pharmacological Therapies in Dementia, p. 4 (2015) ( To appear)Google Scholar
  12. 12.
    Metais, E., Nakache, D., Herradi, N., Ghorbel, F.: Rapport final du projet VIVA: Vivre à Paris avec Alzheimer en 2030 grâce aux nouvelles technologies. Technical report, CEDRIC-14-3246, CEDRIC laboratory, CNAM-Paris, FranceGoogle Scholar
  13. 13.
    Mika, P., Gangemi, A.: Descriptions of social relations. In: 1st Workshop on Friend of a Friend, Social Networking and the (Semantic) Web, September 2004Google Scholar
  14. 14.
    Motik, B., Sattler, U., Studer, R.: Query answering for OWL-DL with rules. J. Web Sem. 3(1), 41–60 (2005)CrossRefGoogle Scholar
  15. 15.
    Sainath, R.: Use of Personal Digital Assistants (PDAs) by Individuals with Acquired Brain Injury: Does it Improve Their Adaptive Function?. Alliant International University, San Diego (2007) Google Scholar
  16. 16.
    Salthouse, T.A.: When does age-related cognitive decline begin? Neurobiol. Aging 30(4), 507–514 (2009)CrossRefGoogle Scholar

Copyright information

© Springer International Publishing Switzerland 2015

Authors and Affiliations

  • Noura Herradi
    • 1
    • 2
  • Fayçal Hamdi
    • 1
  • Elisabeth Métais
    • 1
  • Fatma Ghorbel
    • 3
  • Assia Soukane
    • 2
  1. 1.Cedric LabConservatoire National des Arts Et Métiers (CNAM)ParisFrance
  2. 2.Ecole Centrale d’Electronique (ECE)ParisFrance
  3. 3.Laboratoire MIRACLEUniversité de SfaxSfaxTunisia

Personalised recommendations