DBpedia RDF2Vec Graph Embeddings

DBpedia graph embeddings using RDF2Vec. RDF2Vec embedding generation code can be found here and is based on a publication by Portisch et al. [1]. The embeddings dataset consists of 200-dimensional vectors of DBpedia entities (from 1/9/2021). <strong>Generating Embeddings</strong> The code for generating these embeddings can be found here. Run the run.sh script that wraps all the necessary commmands to generate embeddings <pre><code class="language-bash">bash run.sh</code></pre> The script downloads a set of DBpedia files, which are listed in <code>dbpedia_files.txt</code>. It then builds a Docker image and runs a container of that image that generates the embeddings for the DBpedia graph defined by the DBpedia files. A folder <code>files</code> is created containing all the downloaded DBpedia files, and a folder <code>embeddings/dbpedia</code> is created containing the embeddings in <code>vectors.txt</code> along a set of random walk files. <strong>Run Time of Embeddings Generation</strong> Generating embeddings can take more than a day, but it depends on the number of DBpedia files chosen to be downloaded. Following are some basic run time statistics when embeddings are generated on a 64 GB RAM, 8 cores (AMD EPYC), 1 TB SSD, 1996.221 MHz machine. <strong>Total</strong>: 1 day, 8 hours, 52 minutes, 41 seconds <strong>Walk generation</strong>: 0 days, 7 minutes, 24 minutes, 36 seconds <strong>Training</strong>: 1 day, 1 hour, 28 minutes, 5 seconds <strong>Parameters Used</strong> Here is listed the parameters used to generate the embeddings provided here: <strong>Number of walks per entity</strong>: 100 <strong>Depth (hops) per walk</strong>: 4 <strong>Walk generation mode</strong>: RANDOM_WALKS_DUPLICATE_FREE <strong>Threads</strong>: # of processors / 2 <strong>Training mode</strong>: sg <strong>Embeddings vector dimension</strong>: 200 <strong>Minimum word2vec word count</strong>: 1 <strong>Sample rate</strong>: 0.0 <strong>Training window size</strong>: 5 <strong>Training epochs</strong>: 5