The invention discloses an accurate search recommendation method and
system based on user portraits and
big data analysis, and the method comprises the steps: continuously collecting the attribute information of a promotion activity, constructing an evaluation index, and calculating a potential influence
score of the promotion activity; obtaining the basic weight of the
behavior type, and updating the sensitivity index according to the
behavior type and strength; determining an association relationship between a content item corresponding to the behavior and the promotion activity, if so, calculating an external induction probability based on a potential influence
score and a sensitivity index, and using the external induction probability as an
attenuation factor to update a behavior weight; and generating a recommendation result based on the updated weight, judging the
user feedback after the recommendation result is generated in the time window again, updating and iterating the time accumulation data of the behavior data feedback of the user, and continuously correcting the user portrait and the interest weight thereof, so that the user behavior feedback data under the condition of not neglecting the promotion activity is obtained. And meanwhile, real accurate recommendation is realized.