ld844870209 / kcollectivefp

The source code for our work published in MDM2019, namely k-Collective Influential Facility Placement over Moving Object.

kCollectiveFP

The source code for our work published in MDM2019, namely k-Collective Influential Facility Placement over Moving Object.

Environment

1. These experiments are implementes in C++.

2. IDE is VS 2013.

##Dataset

1. There are two datasets which are recorded by two text files(checkins-10162.txt and checkins-2321.txt).

   • checkins-10162.txt: 10162 users from Gowalla located in California.
   • checkins-2321.txt: 2321 users from Foursquare located in Singapore.

2. The trajectory of user consist of username and a series of check-in points.

3. Candidate sets are randomly generated from check-in points.

4. Both datasets can be found in this page(https://github.com/lihuixidian/PINOCCHIO/tree/master/datasets) ).

##Algorithm

1. PINOCCHIO: We evaluate the inf(·) for all candidates, and select the top k candidates with the maximum inf(·) as the results.
2. GreedyP: The GreedyP algorithm in Algorithm 1 in paper.
3. GreedyPS: The GreedyPS algorithm in Algorithm 2 in paper.

##Supplement

1. PLS.cpp : Main function. 

2. pino.cpp: This file contains the implements of all algorithoms (eg., Baseline, GreedyP, GreedyPS) in our paper.

Usage

1. All data files should be placed in a local folder named as 'Release', e.g., 'D:\Experiment\PLS\Release'.

2. We should load boost library in this project which provides corresponding utilities with respect to R-tree.

3. There are some precompiles in program.

   CANDIDATES_GENERATION : Generate candidates. GEN_FROM_10162 : From 10162 (1) or 2321 (0) dataset. PICK_FROM_UNIQUE : In CANDIDATES_GENERATION, pick from unique coordinates (set) but not all check-in logs (vector). CHECKINS_EXCLUDING: Exclude check-ins that appear in candidates. And generate checkins.txt. DATA_LOADING : Load the datas about candidates and users. They are all in PLS.cpp.

4. HUSH_NUM: the number of bitmaps in pino.h.