PersonalityML 2.0

 

Marca Registrada e Software Registrado

Editores:

 
Jonas Santos Bezerra (DCOMP - Universidade Federal de Sergipe)
Maria Augusta S. N. Nunes (DCOMP - Universidade Federal de Sergipe)
 

1 Introduction 

In the environment of e-commerce the industry is constantly searching for innovative technologies that enable it to treat properly the overload of information, products and services available on the web business. Their main concern is about how to reach personalized and customized offers in order to predict consumer behaviors and then satisfy their expectancy.

The most used technology for treating overload and personalizing information, products and services in e-commerce is recognized as Recommender Systems. Researchers such as Burke proposed at least five techniques for recommendation. Those techniques come to treat the overload problem considering the type of information to be matched towards to personalization. Burke defined them yet as five major techniques: content-based, collaborative filtering, demographic, knowledge-based and utility-based. However, those techniques could not manage adequately the cold-start problem caused mainly by the sparse data and/or new user. This problem was partially solved by the hybrid technique, according to both Burke and Adomavicius and Tuzhilin. A hybrid technique incorporates, at least, two of the already described techniques; usually the collaborative filtering and the content-based techniques are put together.

Even by incorporation of at least two techniques many times the cold-start problem persist yet in recommender systems. Let’s consider the computer in its decision-making process in an overload situation where human efforts could be unable to treat, why computers consider only the information conventionally processed by computer in its decision-making? We mean, take a real situated context such as buying. In that scenario, how the  store’s vendor could partially perceive and extract the consumer behavior and subtle information during his interaction with a client?  The answer is: the vendor may extract clients information’s during his interaction, like (i) what type of clothes the client is using; (ii)  what type of shoes she is using; (iii) is she using watch?; has she a bag?; (iv) Does she use a make up?; (v) is she wearing jewelry?; (vi) how about her haircut?…. And how about another subtle information, like: (i) her speech intonation; (ii) her facial expression; (iii) her gestures; (iv) her body language….  All that information, in a conventional shopping situation, is extracted by the physical/real vendor from a client during their interaction. From that observed data the vendor predicts what type of product the consumer could be interested in.  This type of information has already taken into account by marketing strategies in order to predict costumers’ behavior.  However why that subtle information is not used by computers in its decision-making process?  Is this type of information even collected from user during his interaction with computers? Have they a standard representation in computers?

Subtle information is extremely important for human decision making process, as proved in some studies. They have demonstrated how important psychological aspects of people such as personality traits and emotions are during the human decision-making process.  Even marketing scientists have already described how mandatory this data could be in order to discover the most assertive prediction.

Many subtle information are freely available on web through social networks, blogs, sms, for instance, the question is how to extract them without being too intrusive, and, more importantly, could those subtle information be standardized? Then, how to standardize those data in order to enable to any system receive and use effectively those information as a recommender inputs for web systems.

2 PersonalityML

The Personality markup language (PersonalityML) has the aim of standardize and help to disseminate and share the use of users’ personality information across applications that take human psychological aspects into account in the computer decision making process.

2.1 Document Structure

 

2.1.1 Document Root: The <personalityml>; element

 
Annotation <personalityml>
Definition The root element of a PersonalityML document.
Children The element MUST contain one or more <personality>; elements.
Attributes
  • Required:
    • Namespace declaration for PersonalityML.
  • Optional:
    • any other namespace declarations for application-specific namespaces.
Occurrence This is the root element -- it cannot occur as a child of any other PersonalityML elements.
 

 

 

<personalityml>; is the root element of a document PersonalityML. Involves several elements <personality>; in a single document.

 

The <personalityml>; element needs to define the PersonalityML namespace and can define any others namespaces.

 

Example:

 
<personalityml xmlns="http://www.w3.org/2011/personalityml">;
... 
</personalityml>;

 

2.1.2 Annotation of a single personality, the <personality>; element

 
Annotation <personality>
Definition This element represents a single personality annotation.
Children The only child is optional: <emphasis>; If present it can occur only once
Attributes
  • Required:
    • None for the moment
Occurrence As a child of <personalityml>; or in any markup using PersonalityML
 

The <personality>; element represents a note of an individual personality.

 

 

 

2.2  Personality Representation

 

2.2.1 The <emphasis>; element

 
Annotation <emphasis>
Definition This element represents the emphasis the personality representation belongs to
Children The only child is optional: <approach>
Attributes
  • Required:
    • name, the name of the emphasis
Occurrence As a child of <personality>
 

The <emphasis>; element describes the emphasis on the personality which the annotation belongs, given the value of the attribute name.

 

Example:

 
<personality>;
                <emphasis name="Structure"/>;
</personality>;

 

2.2.2 The <approach>; elemento

 
Annotation <approach>
Definition The approach used for the representation of personality.
Children

All children are optional.

If present, the following child elements can occur only once: <theory>;; <model>;.

Attributes
  • Required:
    • name, indicates the approach.
  • Optional:
    • None for the moment.
Ocurrence As a child of <emphasis>
 

The <approach>; element describes the approach used for the representation of personality.

 

Example:

 
<personality>;
            <emphasis name="Structure">;
                        <approach name="Traits">;
                                    <theory/>;
                         </approach>;
             </emphasis>;
 </personality>;

 

2.2.3 The <model>; element

 
Annotation <model>
Definition The definition of the model.
Children The element MAY contain one <theory>; element.
Attributes
  • Required:
    • name, indicates the name of the model;
  • Optional:
    • None for the moment.
  •  
Ocurrence as a child of <approach>
 

 

 

The <model>; element represents the model being used to represent the personality. The set of values ??that can be used as children is indicated by the name attribute.

 

 

 

2.2.4 The <theory>; element

 
Annotation <theory>;
Definition Theory adopted.
Children

All the children are optional: <inventory>;

If present can occur only one time.

Attributes
  • Required:
    • author, indicates the name the author.
  • Optional:
    • None for the moment.
Ocurrence as a child of <model>; or <approach>;
 

The <theory>; element describes the theory that is being adopted in the note of personality. The author attribute indicates that the author belongs to this theory.

 

 

 

2.2.5 The <inventory>; element

 
Annotation <inventory>;
Definition The definition of the inventory.

Children

It MUST contain one <factors>; element
Attribute
  • Required:
    • test, indicates the inventory.
  • Optional:
    • None for the moment.
Ocurrence as a child of <theory>;
 

The <inventory>; element defines the personality test used to extract the personality.

 

Example:

 
<inventory test="TIPI">;
 

 

 

2.2.6 The <factors>; element

 
Annotation <factors>;
Definition The set of the factors in the inventory
Children It MUST contain one or more <factor>; elements
Attributes
  • Required:
    • None for the moment
  • Optional:
    • set, a name or URI identifying the set of factor names that can be used.
Ocurrence as a child of <inventory>;
 

The <factors>; element defines the set of elements <factor>; which may be used by the attribute value September It should contain at least one element <factor>;.

 

Example:

 
<factors>;
            <factor name="extraversion" score="42"/>; 
</factors>;
 

 

 

2.2.7 The <factor>; element

 
Annotation <factor>;
Definition The definition of the factor (or scale)
Children It MAY contain one <facets>; element
Attributes
  • Required:
    • name, the name of the factor.
  • Optional:
    • score, the (constant) scale value of this factor.
Ocurrence as a child of <factors>;
 

The <factor>; element has the definition of a single personality factor. Element may have a <facets>;.

 
<factors>;
             <factor name="extraversion" score="42">;
             <factor name="agreeableness">;
</factors>;
 

 

 

2.2.8 The <facets>; element

 
Annotation <facets>;
Definition The set of facets for one factor.
Children It MUST contain one or more <facet>; element.
Attributes
  • Required:
    • set, a name or URI identifying the set of facet names that can be used.
Ocurrence as a child of <factor>;
 

The <facets>; element defines the set of facets that a particular factor may have represented the attribute from September <facets>; is an optional child of <factor>;, but when used it is mandatory that has at least one <facet>;.

 

Example:

 
<factor name="extraversion" score="42">;
             <facets set="extraversion facets">;
                         <facet name="friendliness"/>;
             </facets>;
</factor>;
 

 

 

2.2.9 The <facet>; element

 
Annotation <facet>;
Definition The definition of the facet.
Children None
Attributes
  • Required:
    • name, the name of the facet.
  • Optional:
    • score, the (constant) scale value of this facet.
Ocurrence as a child of <facets>;
 

The <facet>; element has an annotation of a single facet of a given factor of personality. The attribute name is required and defines the name of the facet, while the score, which is optional, has a value for this facet.

 

Example:

 
<facets set="extraversion facets">;
            <facet name="friendliness" score="62"/>;
            <facet name="gregariousness" score="44"/>;
            <facet name="assertiveness" score="13"/>;
            <facet name="activity-level" score="46"/>;
            <facet name="excitement-seeking" score="60"/>;
            <facet name="cheerfulness" score="42"/>;
 </facets>;
 

 

 

2.2.10 The <generic>; element

 
Annotation <generic>;
Definition Description of a generic representation of personality.
Children None
Attributes
  • Required:
    • name, the name of the generic element
  • Optional:
    • Score, the (constant) scale value of this element.
Ocurrence A single <generic>; may occur as a child of <theory>;.
 

The <generic>; element provides a generic definition of a representation of personality in which there is a need <model>;.

 

Example:

 
<personality>; 
            <emphasis name="Structure">;
                         <approach name="Traits">;
                                    <theory author="Allport">;
                                                 <generic name="shy"/>;
                                    </theory>;
                         </approach>; 
            </emphasis>; 
</personality>;

2.3 Anotação de Metadados

2.3.1 O elemento <metadata>;

Annotation <metadata>;
Definition The root of the metadata subtree.
Children The element MUST contain one <timestamp>; element, one <place> element and one <tool> element.
Attributes
  • Required
    • none for the moment.
  • Optional:
    • none for the moment.
Ocurrence As a child of <personality>

O elemento <metadata> é a raiz da sub-árvore de metadados que servem para manter informações sobre a forma de extração dos dados, bem como local, língua, usuário e outros dados relevantes à integridade e privacidade das informações psicológicas.

2.3.2 O elemento <timestamp>;

Annotation <timestamp>;
Definition Time-related information
Children The element MUST contain one <date> element;It may contain one <time> element;
Attributes
  • Required
    • none for the moment.
  • Optional:
    • none for the moment.
Ocurrence As a child of <metadata>;

O elemento <timestamp> armazena as informações relacionadas à data e a hora em que os dados psicológicos foram extraídos.

Exemplo:

<metadata>;
<timestamp > ;
...
</timestamp>;
</metadata>;

2.3.3 O elemento <place>;

Annotation <place>;
Definition Information about where the data came from
Children

The element MUST contain one <language> element; It may contain one <country> element;

Attributes
  • Required
    • none for the moment.
  • Optional:
    • none for the moment.
Ocurrence As a child of <metadata>;

O elemento <place>; armazena as informações geográficas sobre de onde vieram os dados, tais como país e língua de origem.

Exemplo:

<metadata>;
<place> ;
...
<place>;
</metadata>;
 

2.3.4 O elemento <tool>;

Annotation <tool>;
Definition Information about the tool and/or methods used to extract the personality data
Children

It may contain one <description> element;

Attributes
  • Requiredname:
    • the name of the tool;
    • type: the type of the tool used.
  • Optional:
    • url: web address where the tool is located (if any);
    • ownerId: attribute for identifying the personality's owner at the tool.
Ocurrence As a child of <metadata>;

O elemento <tool> refere-se à ferramenta que foi utilizada para extrair as informações, seu nome, tipo, como funciona e como pode ser encontrada, dentre outros.

Exemplo:

<tool name="NEO-IPIP 300"
type="Questionary"
ownerId="ID_27"
url="personalityresearch.com.br"> ;
...
</tool>;

2.3.5 O elemento <date>;

Annotation <date>;
Definition The day of the data extraction.
Children

It may contain one <description> element.

Attributes
  • Required name:
    • none for the moment.
  • Optional:
    • none for the moment.
Ocurrence As a child of <timestap>

O elemento <date> refere-se ao dia em que os dados foram extraídos

Exemplo:

<timestamp >;
<date>;2013-01-24</date>;
...
</timestamp>;
 

2.3.6 O elemento <time>;

Annotation <time>;
Definition The hour of the data extraction.
Children

It may contain one <description> element.

Attributes
  • Required name:
    • none for the moment.
  • Optional:
    • none for the moment.
Ocurrence As a child of <timestap>;

O elemento <time> refere-se à hora exata em que os dados foram extraídos.

Exemplo:

<timestamp >;

<time>;14:20:00</time>;
</timestamp>;

 

 

2.3.7 O elemento <county>;

Annotation <country>;
Definition The native country of the owner.
Children

none for the moment;

Attributes
  • Requiredname:
    • none for the moment.
  • Optional:
    • none for the moment.
Ocurrence As a child of <place>

O elemento <time> refere-se à hora exata em que os dados foram extraídos.

Exemplo:

<place>;
<country>;UK</country>;

</place>;

 

2.3.8 O elemento <language>;

Annotation <language>;
Definition The language used in the tool.
Children

none for the moment.

Attributes
  • Required name:
    • none for the moment.
  • Optional:
    • none for the moment.
Ocurrence As a child of <place>;

O elemento <language> armazena a língua em que a ferramenta de extração foi utilizada.

Exemplo:

<place>;

<language>;english</language>;
</place>;
 

2.3.9 O elemento <decription>;

Annotation <description>;
Definition The description of the tool used.
Children

none for the moment.

Attributes
  • Required name:
    • none for the moment.
  • Optional:
    • none for the moment.
Ocurrence As a child of <tool>;

O elemento <description> serve para armazenar uma descrição adicional da ferramenta utilizada.

Exemplo:

<tool … >;
<description>; As 300 questões do NEO-IPIP são divididas igualmente entre os Cinco Grandes Fatores, e cada um dos Fatores se subdivide em seis facetas. Às afirmativas nos itens do inventário, o sujeito respondente atribui um valor de concordância do quanto elas o representam em uma escala tipo Likert de cinco pontos.
</description>/;
</tool>;

 
 
 

References

 

NUNES, M. A. S. N. ; HU, R. . Personality-based Recommender Systems: An Overview . In: ACM Conference on Recommender Systems , 2012, Dublin. Proceedings of the sixth ACM conference on Recommender systems (RecSys '12). New York: ACM, 2012. p. 5-7.

NUNES, M. A. S. N. ; BEZERRA, J. S. ; OLIVEIRA, A. A. . PERSONALITYML: A MARKUP LANGUAGE TO STANDARDIZE THE USER PERSONALITY IN RECOMMENDER SYSTEMS . Revista GEINTEC- Gestão, Inovação e Tecnologias, v. 2, p. 255-273, 2012.

PersonalityML structure

PersonalityML-schema.xsd

NEO-IPIP  PML example

NUNES, M. A. S. N. ; Bezerra, J.S.; Santos, A.C; Oliveira, A. A; Russo, S. L.; Silva, G. F. . PersonalityML.. ENGLISH VERSION. 1. ed. São Cristóvão: UFS, 2012. v. 1. 36 p.