SlideShare a Scribd company logo

Semantics-aware Graph-based Recommender Systems exploiting Linked Open Data

Cataldo Musto, profile picture
Cataldo Musto

The paper discusses semantics-aware graph-based recommender systems that utilize linked open data to enhance recommendation processes. It explores methodologies for utilizing linked open data, including property selection and data modeling, and evaluates their effectiveness through experimental studies using specific datasets. Key research questions focus on the benefits of linked open data features in recommender systems and the impact of feature selection techniques on performance metrics.

Technology
Related topics:
Linked Data
1 of 58
Downloaded 25 times
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
Semantics-aware Graph-based Recommender Systems exploiting Linked Open Data Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis Giovanni Semeraro (Università degli Studi di Bari ‘Aldo Moro’, Italy - SWAP Research Group) UMAP 2016 24th Conference on User Modeling, Adaptation and Personalization Halifax (Canada) July 15, 2016
2 Linked Open Data Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
3 Linked Open Data Methodology to publish, share and link structured data on the Web Definition Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
4 Linked Open Data Cornerstones 1. 2. Use of RDF to publish data on the Web Re-Use of existing properties to express an agreed semantics and connect data sources Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
5 Linked Open Data (cloud) What is it? A (large) set of interconnected semantic datasets Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
6 Linked Open Data (cloud) What kind of datasets? Each bubble is a dataset! Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
7 Linked Open Data (cloud) How many data? 9960 datasets and 149 billions triplessource: http://stats.lod2.eu today! Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
8 Linked Open Data (cloud) DBpedia is the structured RDF mapping of Wikipedia http://dbpedia.org It is the core of the LOD cloud. DBpedia Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
9 Linked Open Data (cloud) Example: unstructured content from Wikipedia example (Wikipedia page) Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
10 Linked Open Data (cloud) How are these data represented? The Matrix Don Davis http://dbpedia.org/resource/Category:Films_shot_in_Australia Films shot in Australia dcterms:subject dbpedia-owl:m usicCom poser http://dbpedia.org/resource/Don_Davis_(composer) dcterms:subject dcterm s:subject dbo:runtimedbpedia-owl:director dcterm s:subject dcterm s:subject Dystopian Films136 American Action Thriller Films Cyberpunk Films The Wachowskis http://dbpedia.org/resource/The_Wachowskis http://dbpedia.org/resource/Dystopian_FIlms http://dbpedia.org/resource/Cyberpunk_Films http://dbpedia.org/resource/American_Action_Thriller_FIlms http://dbpedia.org/resource/Films_About_Rebellions Films about Rebellions Several interesting (non-trivial) features come into play! Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
11 Linked Open Data (cloud) How are these data represented? The Matrix Don Davis http://dbpedia.org/resource/Category:Films_shot_in_Australia Films shot in Australia dcterms:subject dbpedia-owl:m usicCom poser http://dbpedia.org/resource/Don_Davis_(composer) dcterms:subject dcterm s:subject dbo:runtimedbpedia-owl:director dcterm s:subject dcterm s:subject Dystopian Films136 American Action Thriller Films Cyberpunk Films The Wachowskis http://dbpedia.org/resource/The_Wachowskis http://dbpedia.org/resource/Dystopian_FIlms http://dbpedia.org/resource/Cyberpunk_Films http://dbpedia.org/resource/American_Action_Thriller_FIlms http://dbpedia.org/resource/Films_About_Rebellions Films about Rebellions Several interesting (non-trivial) features come into play! Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
12 Research Questions (1) Can we use Linked Open Data for Recommender Systems? Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
13 Research Questions (2) Is it possible to automatically select the most promising properties among those available in the LOD cloud? Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
14 i4 u1 u2 u3 u4 Methodology Graph-based Data Model - original representation Original Graph-based data model Users and Items are connected according to users’ preferences Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
15 i4 u1 u2 u3 u4 Methodology Graph-based Data Model - DBpedia Mapping If we are able to map the items in the dataset with the entities in the LOD cloud, our representation can be extended with new data points Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
16 i4 u1 u2 u3 u4 dcterms:subject http://dbpedia.org/resource/Films_About_Rebellions Films about Rebellions dbprop:director Quentin Tarantino dbprop:director Methodology Graph-based Data Model - LOD-boosted representation 1999 films http://dbpedia.org/resource/1999_films dcterms:subject dcterms:subject Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
17 i4 u1 u2 u3 u4 dcterms:subject http://dbpedia.org/resource/Films_About_Rebellions Films about Rebellions dbprop:director Quentin Tarantino dbprop:director Methodology example - LOD-boosted representation 1999 films http://dbpedia.org/resource/1999_films dcterms:subject dcterms:subject Many new information can be injected in the graph Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
18 i4 u1 u2 u3 u4 dcterms:subject http://dbpedia.org/resource/Films_About_Rebellions Films about Rebellions dbprop:director Quentin Tarantino dbprop:director Methodology example - LOD-boosted representation 1999 films http://dbpedia.org/resource/1999_films dcterms:subject dcterms:subject How to get recommendations? Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
19 Graph-based Recommendation Algorithm Insight: - Calculate PageRank score for each item node. - Sort PageRank scores in a descending order. - Select top-k recommendations PageRank with Priors . T. H. Haveliwala. Topic-Sensitive PageRank: A Context-Sensitive Ranking Algorithm for Web Search. IEEE Trans. Knowl. Data Eng., 15(4): 784–796, 2003. 
 Reference Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
20 i4 u1 u2 u3 u4 dcterms:subject http://dbpedia.org/resource/Films_About_Rebellions Films about Rebellions dbprop:director Quentin Tarantino dbprop:director Methodology example - LOD-boosted representation 1999 films http://dbpedia.org/resource/1999_films dcterms:subject dcterms:subject Is it possibile to automatically select the most promising properties? Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
21 i4 u1 u2 u3 u4 dcterms:subject http://dbpedia.org/resource/Films_About_Rebellions Films about Rebellions dbprop:director Quentin Tarantino dbprop:director Methodology example - LOD-boosted representation 1999 films http://dbpedia.org/resource/1999_films dcterms:subject dcterms:subject X X Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
22 i4 u1 u2 u3 u4 dcterms:subject http://dbpedia.org/resource/Films_About_Rebellions Films about Rebellions dbprop:director Quentin Tarantino dbprop:director Methodology example - LOD-boosted representation 1999 films http://dbpedia.org/resource/1999_films dcterms:subject dcterms:subject X X X Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
23 Experiments Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
24 Research Questions Do graph-based recommender systems benefit of the introduction of LOD-based features? Do graph-based recommender systems exploiting LOD benefit of the adoption of feature selection techniques? 1/2 1. 2. Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
25 Research Questions 3. 4. 2/2 Is there any correlation between the choice of the FS technique and the behavior of the algorithm? (e.g., better diversity or better F1) ? How does our methodology perform with respect to state-of- the-art algorithms? Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
26 Experimental Evaluation Description of the dataset MovieLens 100k 983 users 1,682 movies 100,000 ratings 55.17% positive ratings 84.43 ratings/user (avg.) 48.48 ratings/item (avg.) 93.7% sparsity Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
27 Experimental Evaluation Description of the dataset DBbook dataset 6,181 users 6,733 movies 72,372 ratings 45,85% positive ratings 11.70 ratings/user (avg.) 10.74 ratings/item (avg.) 99.8% sparsity Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
28 Experimental Evaluation Graph Representations :: Recap G Basic Graph with collaborative data points Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
29 Experimental Evaluation GLOD Graph extended with all the properties gathered from the LOD cloud Graph Representations :: Recap Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
30 Experimental Evaluation GLOD+FS Graph encoding only the most relevant properties selected by a feature selection technique FS Graph Representations :: Recap Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
31 Experimental Evaluation Experimental Protocol Algorithm PageRank with Priors Data Split 5-fold Cross Validation for MovieLens Train/Test for DBbook Graph Representation G, GLOD, GLOD+FS Feature Selection Techniques PageRank, Chi-Square, Information Gain, Gain Ratio, mRMR, PCA, SVM #Selected Features top-10, top-30, top-50 properties Evaluation Metrics F1, Intra-List Diversity, Run Time Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
Experiment 1 32 Impact of LOD-based features. Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
F1@5 F1@10 G G_LOD G G_LOD 53 55 57 59 61 60,83 54,24 60,23 53,89 Experiment 1 33 Impact of LOD-based features :: F1-measure Improvement only on MovieLens F1@5 F1@10 G G_LOD G G_LOD 53 56 59 62 65 64,21 55,04 64,31 55,02 MovieLens DBbook Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
Run Time (min.) G G_LOD 50 262,5 475 687,5 900 880 72 Experiment 1 34 Tremendous increase in the run time Impact of LOD-based features :: Run Time Run Time (min.) G G_LOD 50 662,5 1275 1887,5 2500 2.433 100 MovieLens DBbook Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
Experiment 2 35 Impact of Feature Selection techniques Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
Experiment 2 36 GLOD (baseline) = 54,24 Impact of Feature Selection :: MovieLens :: F1@5 Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016 PageRank mRMR Chi-Square SVM Gain Ratio Inf. Gain PCA 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 53 53,5 54 54,5 55 54,31 54,12 54,06 54,21 54,2 54,21 54,12 54,13 53,96 53,98 54,13 54,19 54,29 54,29 54,06 53,97 53,72 53,82 54,14 53,97 54,18
PageRank mRMR Chi-Square SVM Gain Ratio Inf. Gain PCA 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 53 53,5 54 54,5 55 54,31 54,12 54,06 54,21 54,2 54,21 54,12 54,13 53,96 53,98 54,13 54,19 54,29 54,29 54,06 53,97 53,72 53,82 54,14 53,97 54,18 Experiment 2 37 Tree out of seven techniques (and only with 50 features) overcome the baseline Impact of Feature Selection :: MovieLens :: F1@5 Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
PageRank mRMR Chi-Square SVM Gain Ratio Inf. Gain PCA 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 53 53,5 54 54,5 55 54,31 54,12 54,06 54,21 54,2 54,21 54,12 54,13 53,96 53,98 54,13 54,19 54,29 54,29 54,06 53,97 53,72 53,82 54,14 53,97 54,18 Experiment 2 38 Typically, the larger the number of features, the better the F1 #50 #50 #50 #50 #50 #30 #30 (best) Impact of Feature Selection :: MovieLens :: F1@5 Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
PageRank mRMR Chi-Square SVM Gain Ratio Inf. Gain PCA 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 63,5 63,825 64,15 64,475 64,8 64,286 64,19 64,25 64,26 64,19 64,19 64,18 64,32 64,27 64,31 64,3 64,2 64,27 64,22 64,33 64,45 64,35 64,34 64,23 64,35 64,31 Experiment 2 39 #10 #10 #10 #10 #10 #10 Impact of Feature Selection :: DBbook :: F1@10 #10 GLOD (baseline) = 64,20 Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
PageRank mRMR Chi-Square SVM Gain Ratio Inf. Gain PCA 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 63,5 63,825 64,15 64,475 64,8 64,286 64,19 64,25 64,26 64,19 64,19 64,18 64,32 64,27 64,31 64,3 64,2 64,27 64,22 64,33 64,45 64,35 64,34 64,23 64,35 64,31 Experiment 2 40 #10 #10 #10 #10 #10 #10 Impact of Feature Selection :: DBbook :: F1@10 #10 All the techniques overcome the baseline at least once Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
PageRank mRMR Chi-Square SVM Gain Ratio Inf. Gain PCA 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 63,5 63,825 64,15 64,475 64,8 64,286 64,19 64,25 64,26 64,19 64,19 64,18 64,32 64,27 64,31 64,3 64,2 64,27 64,22 64,33 64,45 64,35 64,34 64,23 64,35 64,31 Experiment 2 41 On DBbook best results are obtained with 10 features! #10 #10 #10 #10 #10 #10 (best) Impact of Feature Selection :: DBbook :: F1@10 #10 Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
Run Time (min.) - MovieLens GLOD GLOD+PCA 50 262,5 475 687,5 900 581 880 Experiment 2 42 Significant decrease Impact of Feature Selection techniques :: Run Time Run Time (min.) - DBbook GLOD GLOD+IG 50 687,5 1325 1962,5 2600 1.341 2433 Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
Experiment 3 43 Trade-off between F1 and diversity Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
Experiment 3 44 Trade-off between F1 and diversity Can the choice of the feature selection technique endogenously induce an higher diversity (or, respectively, an higher F1) of the recommendations? Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
Experiment 3 45 Trade-off between F1 and diversity :: MovieLens :: F1@5 G_LOD = Baseline Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
Experiment 3 46 PCA maximizes F1, at the expense of a little diversity Trade-off between F1 and diversity :: MovieLens :: F1@5 Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
Experiment 3 47 Gain Ratio and SVM sacrifice F1, to induce an higher diversity Trade-off between F1 and diversity :: MovieLens :: F1@5 Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
Experiment 3 48 PageRank obtains a good compromise between F1 and Diversity Trade-off between F1 and diversity :: MovieLens :: F1@5 Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
Experiment 3 49 Similar outcomes on DBbook …but more techniques have a good impact Trade-off between F1 and diversity :: MovieLens :: F1@5 Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
Experiment 4 50 Comparison to State of the art BPRMF (Bayesian Personalized Ranking) [+] U2U-KNN (User to User CF) I2I-KNN (Item to Item CF) POPULAR (Popularity-based baseline) [+] S. Rendle, C.Freudenthaler, Z. Gantner, L. Schmidt-Thieme: BPR: Bayesian Personalized Ranking from Implicit Feedback. UAI 2009. Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
Experiment 4 51 Comparison to State of the Art :: MovieLens 50 54 58 62 66 F1@5 F1@10 60,88 54,31 59,16 51,4 59,16 51,78 59,7 52,2 58,35 50,22 I2I-KNN U2U-KNN BPRMF POPULAR PR (G_LOD+PCA) Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
Experiment 4 52 50 54 58 62 66 F1@5 F1@10 60,88 54,31 59,16 51,4 59,16 51,78 59,7 52,2 58,35 50,22 I2I-KNN U2U-KNN BPRMF POPULAR PR (G_LOD+PCA) PageRank with Priors boosted with LOD is the best-performing approach Comparison to State of the Art :: MovieLens Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
Experiment 4 53 50 54 58 62 66 F1@5 F1@10 60,88 54,31 59,16 51,4 59,16 51,78 59,7 52,2 58,35 50,22 I2I-KNN U2U-KNN BPRMF POPULAR PR (G_LOD+PCA) Even state-of-the-art approaches based on Matrix Factorization are overcame by our methodology Comparison to State of the Art :: MovieLens Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
Experiment 4 54 50 55 60 65 70 F1@5 F1@10 64,45 55,4 62,72 52,96 62,63 52,9 62,29 51,93 62,1 51,11 I2I-KNN U2U-KNN BPRMF POPULAR PR (G_LOD+PCA) Behavior confirmed on DBbook Comparison to State of the Art :: DBbook Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
Conclusions 55Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
Recap 56 Methodolology 1. PageRank with Priors as base algorithm 2. Mapping of the items with nodes in the Linked Open Data Cloud 3. Expansion of the data points and injection of new nodes and edges 4. Use of feature selection to automatically select the most promising properties INVESTIGATION ABOUT THE EFFECTIVENESS OF LINKED OPEN DATA IN GRAPH-BASED RECOMMENDER SYSTEMS Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
Lessons Learned 57 Evaluation 1. PageRank with Priors benefit of the injection of data points coming from the LOD cloud 2. Feature Selection techniques improve the results but need to be properly tuned, since its usage is not always useful 3. A significant connection between the choice of the feature selection technique and the maximization of a specific evaluation metric exists 4. PageRank with Priors boosted with LOD significantly overcomes state-of-the-art approaches INVESTIGATION ABOUT THE EFFECTIVENESS OF LINKED OPEN DATA IN GRAPH-BASED RECOMMENDER SYSTEMS Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
thank you Cataldo Musto cataldo.musto@uniba.it @cataldomusto http://www.di.uniba.it/~swap Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016

More Related Content

PDF
Recommender Systems based on Linked Open Data
Cataldo Musto
 
PDF
Tuning Personalized PageRank for Semantics-aware Recommendations based on Lin...
Cataldo Musto
 
PDF
A Deep Architecture for Content-based Recommendations Exploiting Recurrent Ne...
Cataldo Musto
 
PDF
Linked Open Data-enabled Strategies for Top-N Recommendations
Cataldo Musto
 
PDF
Semantics-aware Techniques for Social Media Analysis, User Modeling and Recom...
Cataldo Musto
 
PDF
Mining Big Data and Open Knowledge Sources to develop transparent and serendi...
Cataldo Musto
 
PDF
DeCAT 2015 - International Workshop on Deep Content Analytics Techniques for ...
Cataldo Musto
 
PDF
Semantics-aware Recommender Systems Exploiting Linked Open Data and Graph-bas...
Cataldo Musto
 
Recommender Systems based on Linked Open Data
Cataldo Musto
 
Tuning Personalized PageRank for Semantics-aware Recommendations based on Lin...
Cataldo Musto
 
A Deep Architecture for Content-based Recommendations Exploiting Recurrent Ne...
Cataldo Musto
 
Linked Open Data-enabled Strategies for Top-N Recommendations
Cataldo Musto
 
Semantics-aware Techniques for Social Media Analysis, User Modeling and Recom...
Cataldo Musto
 
Mining Big Data and Open Knowledge Sources to develop transparent and serendi...
Cataldo Musto
 
DeCAT 2015 - International Workshop on Deep Content Analytics Techniques for ...
Cataldo Musto
 
Semantics-aware Recommender Systems Exploiting Linked Open Data and Graph-bas...
Cataldo Musto
 

What's hot (20)

PDF
Automatic Selection of Linked Open Data features in Graph-based Recommender S...
Cataldo Musto
 
PDF
Why should semantic technologies pay more attention to privacy... and vice-ve...
Mathieu d'Aquin
 
PDF
Combining Social Data and Semantic Content Analysis for L’Aquila Social Urban...
Cataldo Musto
 
PDF
Retrieval, Crawling and Fusion of Entity-centric Data on the Web
Stefan Dietze
 
PPT
Combining a co-occurrence-based and a semantic measure for entity linking
Besnik Fetahu
 
PDF
On the many graphs of the Web and the interest of adding their missing links.
Fabien Gandon
 
PDF
Data4Ed - How data sharing, curation and analytics support innovation in educ...
Mathieu d'Aquin
 
PDF
Wimmics Research Team Overview 2017
Fabien Gandon
 
PDF
The Italian Hate Map: semantic content analytics for social good
Cataldo Musto
 
PPTX
Experience from 10 months of University Linked Data
Mathieu d'Aquin
 
PPTX
Thoughts on Knowledge Graphs & Deeper Provenance
Paul Groth
 
PPT
Introduction to question answering for linked data & big data
Andre Freitas
 
PPTX
Working with data.open.ac.uk, the Linked Data Platform of the Open University
Mathieu d'Aquin
 
PPTX
From Research to Innovation: Linked Open Data and Gamification to Design Inte...
Ig Bittencourt
 
PPTX
EKAW2014 Keynote: Ontology Engineering for and by the Masses: are we already ...
Oscar Corcho
 
PDF
Designing a second generation of open data platforms
Yannis Charalabidis
 
PPT
Web 3.0 Emerging
James Hendler
 
PPTX
The Challenge of Deeper Knowledge Graphs for Science
Paul Groth
 
PPTX
Data and science
Anand Deshpande
 
PDF
From Hyperlinks to Semantic Web Properties using Open Knowledge Extraction
STLab
 
Automatic Selection of Linked Open Data features in Graph-based Recommender S...
Cataldo Musto
 
Why should semantic technologies pay more attention to privacy... and vice-ve...
Mathieu d'Aquin
 
Combining Social Data and Semantic Content Analysis for L’Aquila Social Urban...
Cataldo Musto
 
Retrieval, Crawling and Fusion of Entity-centric Data on the Web
Stefan Dietze
 
Combining a co-occurrence-based and a semantic measure for entity linking
Besnik Fetahu
 
On the many graphs of the Web and the interest of adding their missing links.
Fabien Gandon
 
Data4Ed - How data sharing, curation and analytics support innovation in educ...
Mathieu d'Aquin
 
Wimmics Research Team Overview 2017
Fabien Gandon
 
The Italian Hate Map: semantic content analytics for social good
Cataldo Musto
 
Experience from 10 months of University Linked Data
Mathieu d'Aquin
 
Thoughts on Knowledge Graphs & Deeper Provenance
Paul Groth
 
Introduction to question answering for linked data & big data
Andre Freitas
 
Working with data.open.ac.uk, the Linked Data Platform of the Open University
Mathieu d'Aquin
 
From Research to Innovation: Linked Open Data and Gamification to Design Inte...
Ig Bittencourt
 
EKAW2014 Keynote: Ontology Engineering for and by the Masses: are we already ...
Oscar Corcho
 
Designing a second generation of open data platforms
Yannis Charalabidis
 
Web 3.0 Emerging
James Hendler
 
The Challenge of Deeper Knowledge Graphs for Science
Paul Groth
 
Data and science
Anand Deshpande
 
From Hyperlinks to Semantic Web Properties using Open Knowledge Extraction
STLab
 
Ad

Viewers also liked (15)

PDF
La (ir)ragionevole efficacia dei dati
Cataldo Musto
 
PDF
Personalized Wealth Management through Case-based Recommender Systems
Cataldo Musto
 
PDF
Developing Smart Cities Services through Semantic Analysis of Social Streams
Cataldo Musto
 
PDF
Recommender Systems and Linked Open Data
Polytechnic University of Bari
 
PDF
JIST2015-Computing the Semantic Similarity of Resources in DBpedia for Recomm...
GUANGYUAN PIAO
 
PDF
SAC2016-Measuring Semantic Distance for Linked Open Data-enabled Recommender ...
GUANGYUAN PIAO
 
PDF
UMAP2016 - Analyzing Aggregated Semantics-enabled User Modeling on Google+ an...
GUANGYUAN PIAO
 
PDF
SEMANTiCS2016 - Exploring Dynamics and Semantics of User Interests for User ...
GUANGYUAN PIAO
 
PPTX
mu.semte.ch - A journey from TenForce's perspective - SEMANTICS2016
Aad Versteden
 
PDF
EKAW2016 - Interest Representation, Enrichment, Dynamics, and Propagation: A ...
GUANGYUAN PIAO
 
PDF
Case-based Recommender Systems for Personalized Finance Advisory
Cataldo Musto
 
PDF
Building a Recommendation Engine - An example of a product recommendation engine
NYC Predictive Analytics
 
PDF
Recommender system algorithm and architecture
Liang Xiang
 
PDF
Mahout Tutorial and Hands-on (version 2015)
Cataldo Musto
 
PDF
Recommender Systems (Machine Learning Summer School 2014 @ CMU)
Xavier Amatriain
 
La (ir)ragionevole efficacia dei dati
Cataldo Musto
 
Personalized Wealth Management through Case-based Recommender Systems
Cataldo Musto
 
Developing Smart Cities Services through Semantic Analysis of Social Streams
Cataldo Musto
 
Recommender Systems and Linked Open Data
Polytechnic University of Bari
 
JIST2015-Computing the Semantic Similarity of Resources in DBpedia for Recomm...
GUANGYUAN PIAO
 
SAC2016-Measuring Semantic Distance for Linked Open Data-enabled Recommender ...
GUANGYUAN PIAO
 
UMAP2016 - Analyzing Aggregated Semantics-enabled User Modeling on Google+ an...
GUANGYUAN PIAO
 
SEMANTiCS2016 - Exploring Dynamics and Semantics of User Interests for User ...
GUANGYUAN PIAO
 
mu.semte.ch - A journey from TenForce's perspective - SEMANTICS2016
Aad Versteden
 
EKAW2016 - Interest Representation, Enrichment, Dynamics, and Propagation: A ...
GUANGYUAN PIAO
 
Case-based Recommender Systems for Personalized Finance Advisory
Cataldo Musto
 
Building a Recommendation Engine - An example of a product recommendation engine
NYC Predictive Analytics
 
Recommender system algorithm and architecture
Liang Xiang
 
Mahout Tutorial and Hands-on (version 2015)
Cataldo Musto
 
Recommender Systems (Machine Learning Summer School 2014 @ CMU)
Xavier Amatriain
 
Ad

Similar to Semantics-aware Graph-based Recommender Systems exploiting Linked Open Data (20)

PPTX
Self adaptive based natural language interface for disambiguation of
Nurfadhlina Mohd Sharef
 
PPTX
TFF2016, Rudi Studer, Smarte Dienstleistungen mit semantischen Technologien
TourismFastForward
 
PPTX
ESWC 2015 Closing and "General Chair's minute of Madness"
Fabien Gandon
 
PDF
Full Erdmann Ruttenberg Community Approaches to Open Data at Scale
National Information Standards Organization (NISO)
 
PDF
TEAMS 6, 7 and 8
plan4all
 
PPTX
The Future of LOD
Ghislain ATEMEZING
 
PDF
LAK Dataset and Challenge (April 2013)
Stefan Dietze
 
PPT
Linked Data Workshop Stanford University
Talis Consulting
 
PPTX
Streaming HYpothesis REasoning
William Smith
 
PDF
Andy Petrella_Med@Scale by Data Fellas: Scalable and Interoperable Genomics d...
Dataconomy Media
 
PDF
Data Enthusiasts London: Scalable and Interoperable data services. Applied to...
Andy Petrella
 
PPT
PDS Unit - 1 Introdiction to DS.ppt
ssuser52a19e
 
PDF
Knowledge Graph Maintenance
Paul Groth
 
PPTX
Boost your data analytics with open data and public news content
Ontotext
 
PDF
big data analytics introduction chapter 1
terala1
 
PDF
Aioug big data and hadoop
AiougVizagChapter
 
PPTX
Being FAIR: FAIR data and model management SSBSS 2017 Summer School
Carole Goble
 
PDF
Big Data Analytics Lecture notes pdf notes
cseshahinfatima
 
PPTX
Let’s go on a FAIR safari!
Carole Goble
 
PDF
Decentralized Data Management for the Semantic Web
hala Skaf
 
Self adaptive based natural language interface for disambiguation of
Nurfadhlina Mohd Sharef
 
TFF2016, Rudi Studer, Smarte Dienstleistungen mit semantischen Technologien
TourismFastForward
 
ESWC 2015 Closing and "General Chair's minute of Madness"
Fabien Gandon
 
Full Erdmann Ruttenberg Community Approaches to Open Data at Scale
National Information Standards Organization (NISO)
 
TEAMS 6, 7 and 8
plan4all
 
The Future of LOD
Ghislain ATEMEZING
 
LAK Dataset and Challenge (April 2013)
Stefan Dietze
 
Linked Data Workshop Stanford University
Talis Consulting
 
Streaming HYpothesis REasoning
William Smith
 
Andy Petrella_Med@Scale by Data Fellas: Scalable and Interoperable Genomics d...
Dataconomy Media
 
Data Enthusiasts London: Scalable and Interoperable data services. Applied to...
Andy Petrella
 
PDS Unit - 1 Introdiction to DS.ppt
ssuser52a19e
 
Knowledge Graph Maintenance
Paul Groth
 
Boost your data analytics with open data and public news content
Ontotext
 
big data analytics introduction chapter 1
terala1
 
Aioug big data and hadoop
AiougVizagChapter
 
Being FAIR: FAIR data and model management SSBSS 2017 Summer School
Carole Goble
 
Big Data Analytics Lecture notes pdf notes
cseshahinfatima
 
Let’s go on a FAIR safari!
Carole Goble
 
Decentralized Data Management for the Semantic Web
hala Skaf
 

More from Cataldo Musto (20)

PDF
MyrrorBot: a Digital Assistant Based on Holistic User Models for Personalize...
Cataldo Musto
 
PDF
Fairness and Popularity Bias in Recommender Systems: an Empirical Evaluation
Cataldo Musto
 
PDF
Intelligenza Artificiale e Social Media - Monitoraggio della Farnesina e La M...
Cataldo Musto
 
PDF
Exploring the Effects of Natural Language Justifications in Food Recommender ...
Cataldo Musto
 
PDF
Exploiting Distributional Semantics Models for Natural Language Context-aware...
Cataldo Musto
 
PDF
Towards a Knowledge-aware Food Recommender System Exploiting Holistic User Mo...
Cataldo Musto
 
PDF
Towards Queryable User Profiles: Introducing Conversational Agents in a Platf...
Cataldo Musto
 
PDF
Hybrid Semantics aware Recommendations Exploiting Knowledge Graph Embeddings
Cataldo Musto
 
PDF
Natural Language Justifications for Recommender Systems Exploiting Text Summa...
Cataldo Musto
 
PDF
L'IA per l'Empowerment del Cittadino: Hate Map, Myrror, PA Risponde
Cataldo Musto
 
PDF
Explanation Strategies - Advances in Content-based Recommender System
Cataldo Musto
 
PDF
Justifying Recommendations through Aspect-based Sentiment Analysis of Users R...
Cataldo Musto
 
PDF
ExpLOD: un framework per la generazione di spiegazioni per recommender system...
Cataldo Musto
 
PDF
Myrror: una piattaforma per Holistic User Modeling e Quantified Self
Cataldo Musto
 
PDF
Semantic Holistic User Modeling for Personalized Access to Digital Content an...
Cataldo Musto
 
PDF
Holistic User Modeling for Personalized Services in Smart Cities
Cataldo Musto
 
PDF
A Framework for Holistic User Modeling Merging Heterogeneous Digital Footprints
Cataldo Musto
 
PDF
eHealth, mHealth in Otorinolaringoiatria: innovazioni dirompenti o disastrose?
Cataldo Musto
 
PDF
Il Linguaggio dell'Odio sui Social Network
Cataldo Musto
 
PDF
Mappare l'Odio - Hate Speech & Social Media
Cataldo Musto
 
MyrrorBot: a Digital Assistant Based on Holistic User Models for Personalize...
Cataldo Musto
 
Fairness and Popularity Bias in Recommender Systems: an Empirical Evaluation
Cataldo Musto
 
Intelligenza Artificiale e Social Media - Monitoraggio della Farnesina e La M...
Cataldo Musto
 
Exploring the Effects of Natural Language Justifications in Food Recommender ...
Cataldo Musto
 
Exploiting Distributional Semantics Models for Natural Language Context-aware...
Cataldo Musto
 
Towards a Knowledge-aware Food Recommender System Exploiting Holistic User Mo...
Cataldo Musto
 
Towards Queryable User Profiles: Introducing Conversational Agents in a Platf...
Cataldo Musto
 
Hybrid Semantics aware Recommendations Exploiting Knowledge Graph Embeddings
Cataldo Musto
 
Natural Language Justifications for Recommender Systems Exploiting Text Summa...
Cataldo Musto
 
L'IA per l'Empowerment del Cittadino: Hate Map, Myrror, PA Risponde
Cataldo Musto
 
Explanation Strategies - Advances in Content-based Recommender System
Cataldo Musto
 
Justifying Recommendations through Aspect-based Sentiment Analysis of Users R...
Cataldo Musto
 
ExpLOD: un framework per la generazione di spiegazioni per recommender system...
Cataldo Musto
 
Myrror: una piattaforma per Holistic User Modeling e Quantified Self
Cataldo Musto
 
Semantic Holistic User Modeling for Personalized Access to Digital Content an...
Cataldo Musto
 
Holistic User Modeling for Personalized Services in Smart Cities
Cataldo Musto
 
A Framework for Holistic User Modeling Merging Heterogeneous Digital Footprints
Cataldo Musto
 
eHealth, mHealth in Otorinolaringoiatria: innovazioni dirompenti o disastrose?
Cataldo Musto
 
Il Linguaggio dell'Odio sui Social Network
Cataldo Musto
 
Mappare l'Odio - Hate Speech & Social Media
Cataldo Musto
 

Recently uploaded (20)

PDF
Network Security Unit 5.pdf for BCA BBA.
Serpent6
 
PDF
Advanced methodologies resolving dimensionality complications for autism neur...
IAESIJAI
 
PDF
Building Integrated photovoltaic BIPV_UPV.pdf
SandeepSingh286037
 
PDF
Optimiser vos workloads AI/ML sur Amazon EC2 et AWS Graviton
Julien SIMON
 
PPTX
Spectroscopy.pptx food analysis technology
nawahassan45
 
PDF
Encapsulation_ Review paper, used for researhc scholars
gurumoop
 
PPT
“AI and Expert System Decision Support & Business Intelligence Systems”
ZubinRadhakrishnan
 
PDF
Chapter 3 Spatial Domain Image Processing.pdf
Getnet Tigabie Askale -(GM)
 
PPTX
Programs and apps: productivity, graphics, security and other tools
4mqw9zch22
 
PDF
cuic standard and advanced reporting.pdf
SimoneTitaniumTomase
 
PPTX
Effective Security Operations Center (SOC) A Modern, Strategic, and Threat-In...
ReZa AdineH
 
PPTX
VMware vSphere Foundation How to Sell Presentation-Ver1.4-2-14-2024.pptx
blackmambaettijean
 
PDF
The Rise and Fall of 3GPP – Time for a Sabbatical?
3G4G
 
PPT
Teaching material agriculture food technology
LiaRayya
 
PDF
Review of recent advances in non-invasive hemoglobin estimation
IAESIJAI
 
PDF
Unlocking AI with Model Context Protocol (MCP)
Brian McKeiver
 
PDF
Per capita expenditure prediction using model stacking based on satellite ima...
IAESIJAI
 
PDF
Empathic Computing: Creating Shared Understanding
Mark Billinghurst
 
PPTX
ACSFv1EN-58255 AWS Academy Cloud Security Foundations.pptx
23bcla24
 
PDF
NewMind AI Weekly Chronicles - August'25 Week I
NewMind AI
 
Network Security Unit 5.pdf for BCA BBA.
Serpent6
 
Advanced methodologies resolving dimensionality complications for autism neur...
IAESIJAI
 
Building Integrated photovoltaic BIPV_UPV.pdf
SandeepSingh286037
 
Optimiser vos workloads AI/ML sur Amazon EC2 et AWS Graviton
Julien SIMON
 
Spectroscopy.pptx food analysis technology
nawahassan45
 
Encapsulation_ Review paper, used for researhc scholars
gurumoop
 
“AI and Expert System Decision Support & Business Intelligence Systems”
ZubinRadhakrishnan
 
Chapter 3 Spatial Domain Image Processing.pdf
Getnet Tigabie Askale -(GM)
 
Programs and apps: productivity, graphics, security and other tools
4mqw9zch22
 
cuic standard and advanced reporting.pdf
SimoneTitaniumTomase
 
Effective Security Operations Center (SOC) A Modern, Strategic, and Threat-In...
ReZa AdineH
 
VMware vSphere Foundation How to Sell Presentation-Ver1.4-2-14-2024.pptx
blackmambaettijean
 
The Rise and Fall of 3GPP – Time for a Sabbatical?
3G4G
 
Teaching material agriculture food technology
LiaRayya
 
Review of recent advances in non-invasive hemoglobin estimation
IAESIJAI
 
Unlocking AI with Model Context Protocol (MCP)
Brian McKeiver
 
Per capita expenditure prediction using model stacking based on satellite ima...
IAESIJAI
 
Empathic Computing: Creating Shared Understanding
Mark Billinghurst
 
ACSFv1EN-58255 AWS Academy Cloud Security Foundations.pptx
23bcla24
 
NewMind AI Weekly Chronicles - August'25 Week I
NewMind AI
 

Semantics-aware Graph-based Recommender Systems exploiting Linked Open Data

  • 1. Semantics-aware Graph-based Recommender Systems exploiting Linked Open Data Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis Giovanni Semeraro (Università degli Studi di Bari ‘Aldo Moro’, Italy - SWAP Research Group) UMAP 2016 24th Conference on User Modeling, Adaptation and Personalization Halifax (Canada) July 15, 2016
  • 2. 2 Linked Open Data Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 3. 3 Linked Open Data Methodology to publish, share and link structured data on the Web Definition Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 4. 4 Linked Open Data Cornerstones 1. 2. Use of RDF to publish data on the Web Re-Use of existing properties to express an agreed semantics and connect data sources Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 5. 5 Linked Open Data (cloud) What is it? A (large) set of interconnected semantic datasets Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 6. 6 Linked Open Data (cloud) What kind of datasets? Each bubble is a dataset! Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 7. 7 Linked Open Data (cloud) How many data? 9960 datasets and 149 billions triplessource: http://stats.lod2.eu today! Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 8. 8 Linked Open Data (cloud) DBpedia is the structured RDF mapping of Wikipedia http://dbpedia.org It is the core of the LOD cloud. DBpedia Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 9. 9 Linked Open Data (cloud) Example: unstructured content from Wikipedia example (Wikipedia page) Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 10. 10 Linked Open Data (cloud) How are these data represented? The Matrix Don Davis http://dbpedia.org/resource/Category:Films_shot_in_Australia Films shot in Australia dcterms:subject dbpedia-owl:m usicCom poser http://dbpedia.org/resource/Don_Davis_(composer) dcterms:subject dcterm s:subject dbo:runtimedbpedia-owl:director dcterm s:subject dcterm s:subject Dystopian Films136 American Action Thriller Films Cyberpunk Films The Wachowskis http://dbpedia.org/resource/The_Wachowskis http://dbpedia.org/resource/Dystopian_FIlms http://dbpedia.org/resource/Cyberpunk_Films http://dbpedia.org/resource/American_Action_Thriller_FIlms http://dbpedia.org/resource/Films_About_Rebellions Films about Rebellions Several interesting (non-trivial) features come into play! Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 11. 11 Linked Open Data (cloud) How are these data represented? The Matrix Don Davis http://dbpedia.org/resource/Category:Films_shot_in_Australia Films shot in Australia dcterms:subject dbpedia-owl:m usicCom poser http://dbpedia.org/resource/Don_Davis_(composer) dcterms:subject dcterm s:subject dbo:runtimedbpedia-owl:director dcterm s:subject dcterm s:subject Dystopian Films136 American Action Thriller Films Cyberpunk Films The Wachowskis http://dbpedia.org/resource/The_Wachowskis http://dbpedia.org/resource/Dystopian_FIlms http://dbpedia.org/resource/Cyberpunk_Films http://dbpedia.org/resource/American_Action_Thriller_FIlms http://dbpedia.org/resource/Films_About_Rebellions Films about Rebellions Several interesting (non-trivial) features come into play! Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 12. 12 Research Questions (1) Can we use Linked Open Data for Recommender Systems? Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 13. 13 Research Questions (2) Is it possible to automatically select the most promising properties among those available in the LOD cloud? Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 14. 14 i4 u1 u2 u3 u4 Methodology Graph-based Data Model - original representation Original Graph-based data model Users and Items are connected according to users’ preferences Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 15. 15 i4 u1 u2 u3 u4 Methodology Graph-based Data Model - DBpedia Mapping If we are able to map the items in the dataset with the entities in the LOD cloud, our representation can be extended with new data points Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 16. 16 i4 u1 u2 u3 u4 dcterms:subject http://dbpedia.org/resource/Films_About_Rebellions Films about Rebellions dbprop:director Quentin Tarantino dbprop:director Methodology Graph-based Data Model - LOD-boosted representation 1999 films http://dbpedia.org/resource/1999_films dcterms:subject dcterms:subject Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 17. 17 i4 u1 u2 u3 u4 dcterms:subject http://dbpedia.org/resource/Films_About_Rebellions Films about Rebellions dbprop:director Quentin Tarantino dbprop:director Methodology example - LOD-boosted representation 1999 films http://dbpedia.org/resource/1999_films dcterms:subject dcterms:subject Many new information can be injected in the graph Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 18. 18 i4 u1 u2 u3 u4 dcterms:subject http://dbpedia.org/resource/Films_About_Rebellions Films about Rebellions dbprop:director Quentin Tarantino dbprop:director Methodology example - LOD-boosted representation 1999 films http://dbpedia.org/resource/1999_films dcterms:subject dcterms:subject How to get recommendations? Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 19. 19 Graph-based Recommendation Algorithm Insight: - Calculate PageRank score for each item node. - Sort PageRank scores in a descending order. - Select top-k recommendations PageRank with Priors . T. H. Haveliwala. Topic-Sensitive PageRank: A Context-Sensitive Ranking Algorithm for Web Search. IEEE Trans. Knowl. Data Eng., 15(4): 784–796, 2003. 
 Reference Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 20. 20 i4 u1 u2 u3 u4 dcterms:subject http://dbpedia.org/resource/Films_About_Rebellions Films about Rebellions dbprop:director Quentin Tarantino dbprop:director Methodology example - LOD-boosted representation 1999 films http://dbpedia.org/resource/1999_films dcterms:subject dcterms:subject Is it possibile to automatically select the most promising properties? Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 21. 21 i4 u1 u2 u3 u4 dcterms:subject http://dbpedia.org/resource/Films_About_Rebellions Films about Rebellions dbprop:director Quentin Tarantino dbprop:director Methodology example - LOD-boosted representation 1999 films http://dbpedia.org/resource/1999_films dcterms:subject dcterms:subject X X Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 22. 22 i4 u1 u2 u3 u4 dcterms:subject http://dbpedia.org/resource/Films_About_Rebellions Films about Rebellions dbprop:director Quentin Tarantino dbprop:director Methodology example - LOD-boosted representation 1999 films http://dbpedia.org/resource/1999_films dcterms:subject dcterms:subject X X X Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 23. 23 Experiments Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 24. 24 Research Questions Do graph-based recommender systems benefit of the introduction of LOD-based features? Do graph-based recommender systems exploiting LOD benefit of the adoption of feature selection techniques? 1/2 1. 2. Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 25. 25 Research Questions 3. 4. 2/2 Is there any correlation between the choice of the FS technique and the behavior of the algorithm? (e.g., better diversity or better F1) ? How does our methodology perform with respect to state-of- the-art algorithms? Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 26. 26 Experimental Evaluation Description of the dataset MovieLens 100k 983 users 1,682 movies 100,000 ratings 55.17% positive ratings 84.43 ratings/user (avg.) 48.48 ratings/item (avg.) 93.7% sparsity Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 27. 27 Experimental Evaluation Description of the dataset DBbook dataset 6,181 users 6,733 movies 72,372 ratings 45,85% positive ratings 11.70 ratings/user (avg.) 10.74 ratings/item (avg.) 99.8% sparsity Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 28. 28 Experimental Evaluation Graph Representations :: Recap G Basic Graph with collaborative data points Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 29. 29 Experimental Evaluation GLOD Graph extended with all the properties gathered from the LOD cloud Graph Representations :: Recap Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 30. 30 Experimental Evaluation GLOD+FS Graph encoding only the most relevant properties selected by a feature selection technique FS Graph Representations :: Recap Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 31. 31 Experimental Evaluation Experimental Protocol Algorithm PageRank with Priors Data Split 5-fold Cross Validation for MovieLens Train/Test for DBbook Graph Representation G, GLOD, GLOD+FS Feature Selection Techniques PageRank, Chi-Square, Information Gain, Gain Ratio, mRMR, PCA, SVM #Selected Features top-10, top-30, top-50 properties Evaluation Metrics F1, Intra-List Diversity, Run Time Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 32. Experiment 1 32 Impact of LOD-based features. Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 33. F1@5 F1@10 G G_LOD G G_LOD 53 55 57 59 61 60,83 54,24 60,23 53,89 Experiment 1 33 Impact of LOD-based features :: F1-measure Improvement only on MovieLens F1@5 F1@10 G G_LOD G G_LOD 53 56 59 62 65 64,21 55,04 64,31 55,02 MovieLens DBbook Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 34. Run Time (min.) G G_LOD 50 262,5 475 687,5 900 880 72 Experiment 1 34 Tremendous increase in the run time Impact of LOD-based features :: Run Time Run Time (min.) G G_LOD 50 662,5 1275 1887,5 2500 2.433 100 MovieLens DBbook Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 35. Experiment 2 35 Impact of Feature Selection techniques Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 36. Experiment 2 36 GLOD (baseline) = 54,24 Impact of Feature Selection :: MovieLens :: F1@5 Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016 PageRank mRMR Chi-Square SVM Gain Ratio Inf. Gain PCA 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 53 53,5 54 54,5 55 54,31 54,12 54,06 54,21 54,2 54,21 54,12 54,13 53,96 53,98 54,13 54,19 54,29 54,29 54,06 53,97 53,72 53,82 54,14 53,97 54,18
  • 37. PageRank mRMR Chi-Square SVM Gain Ratio Inf. Gain PCA 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 53 53,5 54 54,5 55 54,31 54,12 54,06 54,21 54,2 54,21 54,12 54,13 53,96 53,98 54,13 54,19 54,29 54,29 54,06 53,97 53,72 53,82 54,14 53,97 54,18 Experiment 2 37 Tree out of seven techniques (and only with 50 features) overcome the baseline Impact of Feature Selection :: MovieLens :: F1@5 Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 38. PageRank mRMR Chi-Square SVM Gain Ratio Inf. Gain PCA 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 53 53,5 54 54,5 55 54,31 54,12 54,06 54,21 54,2 54,21 54,12 54,13 53,96 53,98 54,13 54,19 54,29 54,29 54,06 53,97 53,72 53,82 54,14 53,97 54,18 Experiment 2 38 Typically, the larger the number of features, the better the F1 #50 #50 #50 #50 #50 #30 #30 (best) Impact of Feature Selection :: MovieLens :: F1@5 Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 39. PageRank mRMR Chi-Square SVM Gain Ratio Inf. Gain PCA 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 63,5 63,825 64,15 64,475 64,8 64,286 64,19 64,25 64,26 64,19 64,19 64,18 64,32 64,27 64,31 64,3 64,2 64,27 64,22 64,33 64,45 64,35 64,34 64,23 64,35 64,31 Experiment 2 39 #10 #10 #10 #10 #10 #10 Impact of Feature Selection :: DBbook :: F1@10 #10 GLOD (baseline) = 64,20 Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 40. PageRank mRMR Chi-Square SVM Gain Ratio Inf. Gain PCA 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 63,5 63,825 64,15 64,475 64,8 64,286 64,19 64,25 64,26 64,19 64,19 64,18 64,32 64,27 64,31 64,3 64,2 64,27 64,22 64,33 64,45 64,35 64,34 64,23 64,35 64,31 Experiment 2 40 #10 #10 #10 #10 #10 #10 Impact of Feature Selection :: DBbook :: F1@10 #10 All the techniques overcome the baseline at least once Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 41. PageRank mRMR Chi-Square SVM Gain Ratio Inf. Gain PCA 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 10 30 50 63,5 63,825 64,15 64,475 64,8 64,286 64,19 64,25 64,26 64,19 64,19 64,18 64,32 64,27 64,31 64,3 64,2 64,27 64,22 64,33 64,45 64,35 64,34 64,23 64,35 64,31 Experiment 2 41 On DBbook best results are obtained with 10 features! #10 #10 #10 #10 #10 #10 (best) Impact of Feature Selection :: DBbook :: F1@10 #10 Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 42. Run Time (min.) - MovieLens GLOD GLOD+PCA 50 262,5 475 687,5 900 581 880 Experiment 2 42 Significant decrease Impact of Feature Selection techniques :: Run Time Run Time (min.) - DBbook GLOD GLOD+IG 50 687,5 1325 1962,5 2600 1.341 2433 Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 43. Experiment 3 43 Trade-off between F1 and diversity Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 44. Experiment 3 44 Trade-off between F1 and diversity Can the choice of the feature selection technique endogenously induce an higher diversity (or, respectively, an higher F1) of the recommendations? Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 45. Experiment 3 45 Trade-off between F1 and diversity :: MovieLens :: F1@5 G_LOD = Baseline Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 46. Experiment 3 46 PCA maximizes F1, at the expense of a little diversity Trade-off between F1 and diversity :: MovieLens :: F1@5 Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 47. Experiment 3 47 Gain Ratio and SVM sacrifice F1, to induce an higher diversity Trade-off between F1 and diversity :: MovieLens :: F1@5 Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 48. Experiment 3 48 PageRank obtains a good compromise between F1 and Diversity Trade-off between F1 and diversity :: MovieLens :: F1@5 Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 49. Experiment 3 49 Similar outcomes on DBbook …but more techniques have a good impact Trade-off between F1 and diversity :: MovieLens :: F1@5 Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 50. Experiment 4 50 Comparison to State of the art BPRMF (Bayesian Personalized Ranking) [+] U2U-KNN (User to User CF) I2I-KNN (Item to Item CF) POPULAR (Popularity-based baseline) [+] S. Rendle, C.Freudenthaler, Z. Gantner, L. Schmidt-Thieme: BPR: Bayesian Personalized Ranking from Implicit Feedback. UAI 2009. Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 51. Experiment 4 51 Comparison to State of the Art :: MovieLens 50 54 58 62 66 F1@5 F1@10 60,88 54,31 59,16 51,4 59,16 51,78 59,7 52,2 58,35 50,22 I2I-KNN U2U-KNN BPRMF POPULAR PR (G_LOD+PCA) Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 52. Experiment 4 52 50 54 58 62 66 F1@5 F1@10 60,88 54,31 59,16 51,4 59,16 51,78 59,7 52,2 58,35 50,22 I2I-KNN U2U-KNN BPRMF POPULAR PR (G_LOD+PCA) PageRank with Priors boosted with LOD is the best-performing approach Comparison to State of the Art :: MovieLens Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 53. Experiment 4 53 50 54 58 62 66 F1@5 F1@10 60,88 54,31 59,16 51,4 59,16 51,78 59,7 52,2 58,35 50,22 I2I-KNN U2U-KNN BPRMF POPULAR PR (G_LOD+PCA) Even state-of-the-art approaches based on Matrix Factorization are overcame by our methodology Comparison to State of the Art :: MovieLens Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 54. Experiment 4 54 50 55 60 65 70 F1@5 F1@10 64,45 55,4 62,72 52,96 62,63 52,9 62,29 51,93 62,1 51,11 I2I-KNN U2U-KNN BPRMF POPULAR PR (G_LOD+PCA) Behavior confirmed on DBbook Comparison to State of the Art :: DBbook Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 55. Conclusions 55Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 56. Recap 56 Methodolology 1. PageRank with Priors as base algorithm 2. Mapping of the items with nodes in the Linked Open Data Cloud 3. Expansion of the data points and injection of new nodes and edges 4. Use of feature selection to automatically select the most promising properties INVESTIGATION ABOUT THE EFFECTIVENESS OF LINKED OPEN DATA IN GRAPH-BASED RECOMMENDER SYSTEMS Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 57. Lessons Learned 57 Evaluation 1. PageRank with Priors benefit of the injection of data points coming from the LOD cloud 2. Feature Selection techniques improve the results but need to be properly tuned, since its usage is not always useful 3. A significant connection between the choice of the feature selection technique and the maximization of a specific evaluation metric exists 4. PageRank with Priors boosted with LOD significantly overcomes state-of-the-art approaches INVESTIGATION ABOUT THE EFFECTIVENESS OF LINKED OPEN DATA IN GRAPH-BASED RECOMMENDER SYSTEMS Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016
  • 58. thank you Cataldo Musto cataldo.musto@uniba.it @cataldomusto http://www.di.uniba.it/~swap Cataldo Musto, Pasquale Lops, Pierpaolo Basile, Marco de Gemmis, Giovanni Semeraro Semantics-aware Graph-based Recommender Systems Exploiting Linked Open Data. UMAP 2016, Halifax (Canada). 15.07.2016