PROJECT TITLE:

Personalized Travel Sequence Recommendation on Multi-Source Big Social Media - 2016

ABSTRACT:

Huge information increasingly benefit both analysis and industrial space like health care, finance service and commercial recommendation. This paper presents a personalised travel sequence recommendation from each travelogues and community contributed photos and also the heterogeneous metadata (e.g., tags, geo-location, and date taken) associated with these photos. Unlike most existing travel recommendation approaches, our approach isn't only personalized to user's travel interest but conjointly in a position to advocate a travel sequence rather than individual Points of Interest (POIs). Topical package house together with representative tags, the distributions of price, visiting time and visiting season of every topic, is mined to bridge the vocabulary gap between user travel preference and travel routes. We tend to use the complementary of two kinds of social media: travelogue and community contributed photos. We tend to map each user's and routes' textual descriptions to the topical package house to urge user topical package model and route topical package model (i.e., topical interest, value, time and season). To recommend personalised POI sequence, initial, famous routes are ranked in keeping with the similarity between user package and route package. Then high ranked routes are additional optimized by social similar users' travel records. Representative images with viewpoint and seasonal diversity of POIs are shown to supply a more comprehensive impression. We evaluate our recommendation system on a assortment of 7 million Flickr pictures uploaded by 7,387 users and 24,008 travelogues covering 864 travel POIs in 9 famous cities, and show its effectiveness. We additionally contribute a replacement dataset with additional than 200 K photos with heterogeneous metadata in nine famous cities.