Match and Meet Services

De Lillois Fractale Wiki
Aller à : navigation, rechercher

Introduction

This is one project among others from PG.

This document only presents a modelling approach. assumed to be different and better compared to the underlying model of existing meeting web sites.

However the most critical and most difficult tasks will be elsewhere:

  • find an initial profile base (buy it,...)
  • define a revenue model
  • organize a payment model and process
  • market the new service, advertize for it
  • set up a company structure
  • make it sellable to major web actors
  • ...


Model overview

Modelling principle

The existing web sites poorly replicates the subtle nature of human affinity, meeting process, matching issues,...

This model is an attempt to better match that subtle nature.

Success of the model

The model is successful if...

  • effective matching occurs
  • matched members are indded suited to each other
  • revealing process is correctly and suitably used
  • members are willing to fill profile
  • members are willling to folloow the matching process as expected

Fuzziness

The model uses fuzzy logic, better than the naive binary logic used in most matching services.

Alchemy

The model takes in account the unpredictable affinity perceived by interacting human beings.

Underlying model and user experience

The underlying model contains various more or less complex concepts.

It is important to translate these complex concepts into intuitive user interface and user experience. The user interface sould be readable by members with average cultural level.

Basic concepts

Member. A member is an individual willing to meet matching profiles.

Community. A community is a set of members. All members of a community are visible by all other members of the same community, at least at the lowest revealing level.

Data. Information related to a member, provided by himself. Many data are optional (they may not be set). Data are either general data or media data.

Fit. Information related to 2 combined members. Obviously, if the community contains N members, then the number of fits will be potentially Nx(N-1). It is a matrix. Note that the matrix is not symmetrical. Fij is the fit of member i on member j's mind, while Fij is the fit of member i on member j's mind.

Advanced Concepts

Confidentiality level

Any data related to a member receives a confidentaility level. Depending on the data, the confidentiality level is determined by the member introducing his own profile.

The most private data has confidentiality level C9.

The most public data has confidentiality level C1.

It is assumed here that 9 confidentiality levels are available, but this is tunable.

Revealing level

For some data, the member is allowed to specify a reveal level. This level decided by the member is a level, on the same scale as the confidentiality level.

For other data, the revealing level may not be changed. This is true for the sex, the age, the pseudo,... These data have a confidentiality level of C1.

Reveal matching

The view of member i made available to member j is decided by member i. It is a level, on the same scale as the confidentiality level.

Initially the revealing levels of any pair of member is C1. But at any time, a member is able to modify(increase or decrease):

  • the default revealing level given to all other members (at least C1).
  • the revealing level given to one identified member (C1 to C9).

This is the modelling echo of the human process, where people progressively decide to give more information about themselves to chosen potential partners.

Personal general data

This is a set of data that has to be set for any member, with level C1.

  • Chosen pseudo
  • Sex
  • Matching sex (searched)
  • Age (actually date of birth)
  • Size
  • Weight
  • Mother langage
  • Contact langages
  • Location (low precision : department in France, province in Belgium,...)
  • [ auto generated simple text ]

Personal private data

This is a set of data that may be filled or not, and for which the user is allowed to tune the revealing level:

  • self portrait (with C1 title; as many as desired, with possibly different Cx)
  • free describing texts (with C1 title; as many as desired, with possibly different Cx)
  • pictures (with C1 title; as many as desired, with possibly different Cx)
  • voice messages (with C1 title; as many as desired, with possibly different Cx)
  • video (with C1 title; as many as desired, with possibly different Cx)
  • location (as accurate as desired)
  • smoking habits
  • hair colour
  • eye colour
  • sport affinities (list of)
  • food affinities (list of)
  • hobby affinities (list of)
  • preferred reading (list of)
  • preferred movies (list of)
  • professional category
  • professional activity
  • religion
  • racial type
  • marital status
  • family / children situation
  • attached animals
  • personal specifics
  • nature (attributes)

Searched profile

All members specify their target partners with more or less details. The target partner is defined as a searched profile, with the following attributes:

  • optimal size (+ range)
  • optimal weight (+ range)
  • optimal age (+ range)
  • hair colour
  • eye colour
  • smoking habits
  • distance
  • professional category
  • professional activity
  • religion
  • racial type
  • marital status
  • family / children situation
  • contact langage
  • matching sports
  • matching hobbies

Criterium weight

All components of the searched profile receive a weight decided by the searching member.

The weight is defined on a scale ranging from W0 (irrelevant criterium) to W9 (most important criterium). So a member my decide that smoking habits are extremely important or absutely irrelevant, and so the the age, langage,...

Subjective affinity

In the process of getting to know other members, any member will perceive a subjective affinity, wich will be tuned up and down when

  • discovering revealed attributes
  • exchanging messages
  • chatting
  • meeting in the real world

The member will be given the opportunity to define (and update at any time) this subjective affinity factor. It is called A-value and it is a matrix. The matrix is asymmetric.

Matching score

For any pair of member, a matching score is computed. The computation is a straightforward combination of

  • searched profile data
  • target member data
  • criterium weights
  • subjective affinity

The matching score (M-value) is organized as a matrix. The matrix is asymmetric, i.e. it contains different values for Mij and Mji.

Fitting score

The fitting score is the symmetric combination (average) of a pair of matching scores. The fitting score (F-value) is organized as a symmetric matrix. Fij = Jji = (Mij + Mji)/2

Matching market map

For a given member, the set of all other members may be seen as 2-dimension map, based on the matching scores. For example, it may be represented in a map with points distributed from left (I do not want him/her) to right (I want him/her), and from bottom (he/she does not want me) to top (he/she wants me).

Reveal process

The meeting process is mainly a reveal process. On this the model is similar to the real world meeting rules. One can imagine that after having defined a number of private data (more or less private data), the member decides to give more and more access to priviledged partners on progressively more privata data.

Journal

In a match and meet community, complex events are likely to happen in a wide time sacle between numerous individuals. For example:

some members are sometimes "out of the market" because they have found one identified partner, but later come back to the service when the relationship has ended for some reason

some members find numerous prospects and do not have the time and opportunity to quickly review potentially interesting companions

some members evolve, because their situation, location, habits change in the course of time

For all these reasons, it is useful for any member to have a trace keeping mechanism (a journal) for all members interacting with him/her. The journal is a matrix a events occuring between a pair a members. Any member may decide to erase some of his/her past relationship journal.

The journal contains

  • visit history
  • messages
  • chatting track
  • personal notes (written at any time)
  • revealing level evolutions

User interface concepts

For some parts, the user interface is similar to what is currently seen

  • part "my profile"
  • part "searched profile"
  • messaging service
  • chatting service
  • exploring tools

Some specific parts:

  • Reveal autorization message
  • Reveal request message
  • Visit message
  • Chat request message
  • Chat answer message
  • Notes management

But the most important part is a set of "inspiring" views:

  • matching pair (you and me)
  • market map

Technology

PG's preferences would be Java (applet or applic or JNLP). Fat client. More versatile, reacting. Demanding more local power. Requiring strong JVM devices.

Other solution would be web applications, with light clients. More widely available. Less versatile.



Modèle:XT