1) Geographical Knowledge (GeoKB). This approach This paper describes our Georeferencing approaches, experi- was used for MediaEval 2010 Placing Task [3] and then was ments, and results at the MediaEval 2011 Placing Task eval- improved (see [4]). The GeoKB approach uses the Geonuation. The task consists of predicting the most probable ames1 Gazetteer for detecting the place names, stopwords geographical coordinates of Flickr videos. Our approaches lists, and an English Dictionary. The system uses the folused only Flickr users textual annotations and tagsets to pre- lowing rules from Toponym Disambiguation techniques [4] dict. We used three approaches for this task: 1) a Geograph- to get the geographical focus of the video: 1) select the most ical Knowledge approach, 2) an Information Retrieval based populated place that is not a state, country or continent and approach with Re-Ranking, and 3) a combination of both has its state apearing in the text, 2) otherwise select the (GeoFusion). The GeoFusion approach achieved the best most populated place that is not a state, country or contiresults within the margin of errors from 10km to 10000km. nent and has its country apearing in the text, 3) otherwise select the most populated state that has its country apearing in the text 4) otherwise apply population heuristics. Categories and Subject Descriptors 2) Information Retrieval with Re-Ranking. This H.3 [Information Search and Retrieval] approach is similar to the one presented by [6]. It uses the Terrier2 IR software (version 3.0) with the Hiemstra LanGeneral Terms gduefaagueltMlaomdebldliang( ()HLpaMra)mweetiegrhtvianlguemiondTeler[r5i]e.r (T0h.1e5)HwLMas Design, Performance, Experimentation, Measurement used. See in equation 1 the Terrier implementation of the HLM Weighting model (version 1 [5]) score of a term t in
document length.
The MediaEval 2011 Placing task requires that
participants automatically assign geographical coordinates
(latitude and longitude) to Flickr videos using one or more of:
Flickr metadata, visual content, audio content, and social
information (see [
Score(t; d) = log (1 + tft;d Pi cfi ) (eq.1 ) (1 ) cft Pi tfi;d
The indexing of the metadata subsets were done with the coordinates as a document number and their associated tagsets as the document text. We indexed with ltering using the multilingual stopwords list and without stemming. The following metadata elds (lowercased) from the videos were used for the query: Keywords (tags), Title and Description. A Re-Ranking (RR) process is applied after the IR process. For each topic their rst 1000 retrieved coordinates pairs from the IR software are used. From them we selected the subset of coordinates pairs with a weight equal or greater than the two-thirds (66.66%) of the weight of the coordinates pair ranked in rst position. Then for each geographical coordinates pair of the subset we sum its associated weight (provided by the IR software) and the weight of their neighbours at a threshold distance (e.g. 100km). Then we select the one with the maximum weighted sum. 1Geonames. http://www.geonames.org 2Terrier. http://terrier.org 3) Combination of GeoKB and Information Retrieval with Re-Ranking (GeoFusion). The GeoFusion approach is applied by combining the results of the GeoKB approach and the IR approach with Re-Ranking. From the GeoKB system are selected the predicted coordinates that come only from the Geographical Knowledge heuristics 1, 2 and 3 (avoiding predictions from the population heuristics rules). When the GeoKB rules (applied in priority order: 1, 2, and 3) do not match then the predictions are selected from the IR approach with Re-Ranking.
We used two corpora for training the IR system for
MediaEval 2011: 1) the MediaEval 2011 Flickr corpus (3,185,258
photos) and 2) the union of the MediaEval corpus with
the CoPhIR3 image collection [
We designed a set of four experiments (see Table 1) for the MediaEval 2011 Placing Task test set of 5347 Flickr videos. The experiment TALP1 used the IR approach with Re-Ranking up to 100 km and the MediaEval 2011 photos corpu as a training data. The experiment TALP2 used the GeoKB approach. The experiment TALP3 used the GeoFusion approach with the MediaEval training corpora. The experiment TALP5 used the GeoFusion approach with the MediaEval and the CoPhIR corpora of photos for training. The results are shown in Figure 1 and Table 2.
Approach IR Re-Rank (100km)
GeoKB GeoKB + IR Re-Rank (100km) GeoKB + IR Re-Rank (100km) Training Corpus
MediaEval
MediaEval MediaEval+ CoPhIR
We used three approaches at MediaEval 2011 Placing Task. The GeoFusion approach achieved the best results in the experiments clearly outperforming the other approaches. This 3CoPhIR. http://cophir.isti.cnr.it 40 % 30 % 20 % 10 % approach achieves the best results because combines high precision rules based on Toponym Disambiguation heuristics and predictions that come from a data driven IR ReRanking approach. The GeoKB rules used in the GeoFusion approach achieved 80.18% of accuracy (1789 of 2231 videos) predicting up to 100km. As a further work we plan to improve the accuracy of the GeoKB rules.
This work has been supported by the Spanish Research Dept. (KNOW 2, TIN2009-14715-C04-04). Daniel Ferres is supported by the EBW II Project, which is nanced by the European Commission within the framework of the Erasmus Mundus Programme. TALP Research Center is recognized as a Quality Research Group (2001 SGR 00254) by DURSI, the Research Department of the Catalan Government.