DISCLAIMER:

The data set and the code segments are freely available, for as long as they are not used for research (not commercial) purposes. Everything is provided "AS IS", with no warranty or additional support (we may update the data if/when we continue the next step of the research and are under no obligation to make individual notifications). You can think of it as a version of BSD type of a licence, whatever -- you know what we mean...

Since we did use publicly available sources:

1.a MNTG (Minnesota Network Traffic Generator): http://mntg.cs.umn.edu/tg/index.php

1.b Open Street Map: https://www.openstreetmap.org

1.c Several (three in the SSTD 2015 submission) different similarity measures based on WordNet, obtained using: http://maraca.d.umn.edu/

we recommend (out of a courtesy, if nothing else) that you cite the respective sources if/when using our data. 

Note, however, that the data from each of those sources was respectivelly modified as follows:

2.a The MNTG trajectories were augmented with POIs (Points of Interest) and were "injected" with duration of stay at such locations.

2.b The OMS data was used to generate the boundaries for the zones in the city of Chicago (used in the experiments).

2.c Both Datasets were uploaded in the SQL Server 2012 using appropriate scripts, and were used for generating experimental observations, relying on a table populated with values for the three similarity measures (Leacock & Chodorow; Wu & Palmer; Resnik) obtained via "1.c" above".

Lastly, the folder which contains this file also contains 3 .zip files, each corresponding to:  

3.a Trajectories-Regular_and_Semantics: we have three types of files here -- original MNTG outcoumes; MNTG with POIs injected; and POIs-only semantic trajectories. We note that the total size is in the order of 10s of GB and we'd be happy to share the google link with anyone interested (we did generate different amount of trajecotires per-run, and with different lengths...). In this folder, we have provided 3 sub-folders ("1000v", "2000v" and "3000v") as samples, with a total of 3 x 48 = 144 files. We also provide the java code for generating the semantic trajectories out of the "raw" trajectories + POIs as input, as well as another "Readme" file;

Note: the "naming convention is such that a folder with name a000vb000t denotes that there were a total "a-thousand" trajectories, with a max.-duration of "b-thousand" time-units (a time-unit is approx. 2 seconds) generated from that particular run of the MNTG simulator, subsequently converted to semantic trajectories...


3.b POIs-Zones-Similarity: here we provide the locations/facilities, zones and and the java sources for generating similarity-values based on the 3 measures and  WordNet;

3.c. SQL Stuff: All the SQL Server related stuff (upload-scripts; SQL queries);

Inside each, there is brief context-specific readme. 
 