Practical Application of Algorithms

Now that you have familiarized yourself with graph algorithms and how to execute them in the Graph Data Science library, you will do a quick graph analysis of the Recipes dataset.

First, you will infer a similarity network between ingredients with the help of the Node Similarity algorithm. The algorithm will compare how often the ingredients are used together in the same recipe. As in the previous lesson, you will start by examining the Jaccard Similarity score distribution to help you define reasonable topK and similarityCutoff parameters. This way, the inferred network between ingredients will not be too sparse or too dense.

Next, you will use the Louvain Modularity algorithm to analyze the community structure of the ingredient similarity network. You will investigate how the topK and similarityCutoff parameters of the Node Similarity algorithm directly influence the size and the structure of communities.

Finally, you will use the PageRank algorithm to find the representatives of those communities. When using the PageRank algorithm, think of every relationship as a vote of confidence or influence. In our analysis, every relationship is a vote of commonly used together ingredients, and so, the most frequently used together ingredients will rank the highest. You could think of them as the representatives of those communities.