compared with
Current by boyan.kukushev
on Nov 20, 2014 16:57.

Key
This line was removed.
This word was removed. This word was added.
This line was added.

Changes (14)

View Page History
The recommendation engine has a lot of internal parameters for fine tuning the underlying algorithm. For example, when determining the relevance of any given article, its popularity and freshness can alter its score (popular and recent articles score higher). The weight of popularity and freshness when computing the article score can be controlled by setting the respective parameters ({{beh.weight.popularity}} and {{beh.weight.freshness}} respectively) to better suit particular needs.
the relevance of any given article, its popularity and freshness can alter its score (popular and recent articles score higher).
The weight of popularity and freshness when computing the article score can be controlled by setting the respective parameters
({{beh.weight.popularity}} and {{beh.weight.freshness}} respectively) to better suit particular needs. There are many more parameters
to adjust, see the reference at the end of this document.

Parameters can be set per individual request or globally, for all requests
There are many more parameters to adjust, see the reference at the end of this document.

Parameters can be set per individual request or globally, for all requests.

h3. Passing parameters for recommendation request

Both {{/recommend/contextual}} and {{/recommend/behavioural}} calls can be called with {{GET}} and {{POST}} HTTP methods. When {{POST}}-ing, some advanced parameters could be passed for fine tuning the results. {{Content-type}} for {{POST}} requests should be *{{application/json}}* and the request body should contain a simple JSON object with parameter name and value pairs. For example:
{code}
{code:javascript}
POST /recommend/contextual?contentid=<existing_id>
Content-type: application/json

All recognized parameters, values below are the defaults:

{code:javascript}
[
"type": "DOUBLE",
"value": 1.0,
"description": "the relative weight of the freshness/recency of an article (only when beh.boost.expr is missing)"
},
{
"type": "DOUBLE",
"value": 1.0,
"description": "the relative weight of an article's popularity (only when beh.boost.expr is missing)"
},
{
"type": "DOUBLE",
"value": 1.0,
"description": "the relative weight of article covisitation (only when beh.boost.expr is missing)"
},
{
"type": "DOUBLE",
"value": 1.0,
"description": "the weight of the article content similarity portion when calculating behavioural recommendation"
},
{
"type": "DOUBLE",
"value": 1.0,
"description": "the weight of the user profile similarity portion when calculating behavioural recommendation"
},
{
"type": "STRING",
"value": "title,summary,content,tags,keyphrases",
"description": "the fields to use for content similarity"
},
{