Recommendation’s Engine based on Spread Activation algorithm

Suggestion graph

Suggestion graph made with test application

Hi,

Since last year that I haven’t added any post on my blog,  but I would like add new posts. This year I’ll start sharing publishing the work carried out in order to get my degree. The document is currently in spanish.

Here is the abstract:

Nowadays people are reproducing the social network from your real life
into a virtual space in which are represented the same social structures
and relations of friendship, work, academic partners, and “love-
relationships”.
This thesis focuses on the latter type of relations, in order to investigate,
create and analyze a “recommendation engine” capable of generating
recommendations of “couples-friend” to a registered couple in a site
designed for them, where only can sign-in couples previously formed.
For the process of development was used “Research Methodology” and
the software was made using methodology “Extreme Programming”
The recommend engine is based on relationships with other couples
registered with the site. Relationships are represented as graphs where
nodes symbolize each couple and the ties between nodes represent
friendship between them. To carry out the suggestions of the way of: “You
might know this couple …” were used and compared two algorithms. The
first of them is based on the number of nodes (pairs) in common and
another in the adaptation of a search method used in semantic network
which is know as “Spread Activation” in which, from the couple being
analyzed, an amount of energy is propagated to adjacent couples.
In order to capture test data was developed a Facebook Application using
the Python programming language together with the frame-work for
Django web development, and other tools.
Finally it was concluded, using the performance measures know as
“Precision” and “Recall”, that the suggestions that were based on
“Spreading Activation” are slightly better than those based only on the
number of mutual friends.

Read full document

Regards,

Álvaro

Advertisements

8 responses to “Recommendation’s Engine based on Spread Activation algorithm

  • Marcel

    Dear Alvaro, I am Brazillian M.S.C student and I am working on this field. I’ve found your post and it is quite interesting your work. But the link for your thesis is not working. How can I get access for your thesis ?
    I am working with a social network dataset and I am planning to add this new functionality of recommending new users based on their social relationships.

    Congratulations for your work,

    Regards,

    Marcel

  • Marcel

    Thanks Alvaro! Please if it is possible inform here at the comments when the link is already available.

  • Marko

    Hi Alvaro,

    I am also interested in your work, I am a PhD student from Serbia. Could you please put your work online again?

    Thank you

    Marko

    • abrange

      Hi Marko,
      I’ll upload again document when arrive at home.
      Anyway, I would like know In what are you working in your phd?

      Regards,
      Álvaro Brange
      @abrange

  • aleksendric

    Hi Alvaro, thanks for the fast reply. My main field of interest are the control systems, singular, time dalayed, but also some robust stability problems. I find these algorithms, though coming from a whole different field, very interesting to say the least.

    • abrange

      Aleksendric.
      I just update the link with the document. I would like get some feedback if you read the document.
      thank you,
      Álvaro

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: