diff options
Diffstat (limited to 'docs/posts')
| -rw-r--r-- | docs/posts/2022-05-21-Similar-Movies-Recommender.html | 438 | ||||
| -rw-r--r-- | docs/posts/index.html | 17 |
2 files changed, 455 insertions, 0 deletions
diff --git a/docs/posts/2022-05-21-Similar-Movies-Recommender.html b/docs/posts/2022-05-21-Similar-Movies-Recommender.html new file mode 100644 index 0000000..42b887a --- /dev/null +++ b/docs/posts/2022-05-21-Similar-Movies-Recommender.html @@ -0,0 +1,438 @@ +<!DOCTYPE html> +<html lang="en"> +<head> + + <link rel="stylesheet" href="/assets/main.css" /> + <link rel="stylesheet" href="/assets/sakura.css" /> + <meta charset="utf-8"> + <meta name="viewport" content="width=device-width, initial-scale=1.0"> + <title>Hey - Post - Building a Simple Similar Movies Recommender System</title> + <meta name="og:site_name" content="Navan Chauhan" /> + <link rel="canonical" href="https://web.navan.dev/" /> + <meta name="twitter:url" content="https://web.navan.dev/" /> + <meta name="og:url" content="https://web.navan.dev/" /> + <meta name="twitter:title" content="Hey - Post - Building a Simple Similar Movies Recommender System" /> + <meta name="og:title" content="Hey - Post - Building a Simple Similar Movies Recommender System" /> + <meta name="description" content=" Building a Content Based Similar Movies Recommender System " /> + <meta name="twitter:description" content=" Building a Content Based Similar Movies Recommender System " /> + <meta name="og:description" content=" Building a Content Based Similar Movies Recommender System " /> + <meta name="twitter:card" content=" Building a Content Based Similar Movies Recommender System " /> + <meta name="viewport" content="width=device-width, initial-scale=1.0" /> + <link rel="shortcut icon" href="/images/favicon.png" type="image/png" /> + <link rel="alternate" href="/feed.rss" type="application/rss+xml" title="Subscribe to Navan Chauhan" /> + <meta name="twitter:image" content="https://web.navan.dev/images/logo.png" /> + <meta name="og:image" content="https://web.navan.dev/images/logo.png" /> + <link rel="manifest" href="manifest.json" /> + <meta name="google-site-verification" content="LVeSZxz-QskhbEjHxOi7-BM5dDxTg53x2TwrjFxfL0k" /> + <script async src="//gc.zgo.at/count.js" data-goatcounter="https://navanchauhan.goatcounter.com/count"></script> + +</head> +<body> + <nav style="display: block;"> +| +<a href="/">home</a> | +<a href="/about/">about/links</a> | +<a href="/posts/">posts</a> | +<a href="/publications/">publications</a> | +<a href="/repo/">iOS repo</a> | +<a href="/feed.rss">RSS Feed</a> | +</nav> + +<main> + <h1>Building a Simple Similar Movies Recommender System</h1> + +<h2>Why?</h2> + +<p>I recently came across a movie/tv-show recommender, <a rel="noopener" target="_blank" href="https://couchmoney.tv/">couchmoney.tv</a>. I loved it. I decided that I wanted to build something similar, so I could tinker with it as much as I wanted.</p> + +<p>I also wanted a recommendation system I could use via a REST API. Although I have not included that part in this post, I did eventually create it.</p> + +<h2>How?</h2> + +<p>By measuring the cosine of the angle between two vectors, you can get a value in the range [0,1] with 0 meaning no similarity. Now, if we find a way to represent information about movies as a vector, we can use cosine similarity as a metric to find similar movies.</p> + +<p>As we are recommending just based on the content of the movies, this is called a content based recommendation system.</p> + +<h2>Data Collection</h2> + +<p>Trakt exposes a nice API to search for movies/tv-shows. To access the API, you first need to get an API key (the Trakt ID you get when you create a new application). </p> + +<p>I decided to use SQL-Alchemy with a SQLite backend just to make my life easier if I decided on switching to Postgres anytime I felt like. </p> + +<p>First, I needed to check the total number of records in Trakt’s database.</p> + +<div class="codehilite"><pre><span></span><code><span class="kn">import</span> <span class="nn">requests</span> +<span class="kn">import</span> <span class="nn">os</span> + +<span class="n">trakt_id</span> <span class="o">=</span> <span class="n">os</span><span class="o">.</span><span class="n">getenv</span><span class="p">(</span><span class="s2">"TRAKT_ID"</span><span class="p">)</span> + +<span class="n">api_base</span> <span class="o">=</span> <span class="s2">"https://api.trakt.tv"</span> + +<span class="n">headers</span> <span class="o">=</span> <span class="p">{</span> + <span class="s2">"Content-Type"</span><span class="p">:</span> <span class="s2">"application/json"</span><span class="p">,</span> + <span class="s2">"trakt-api-version"</span><span class="p">:</span> <span class="s2">"2"</span><span class="p">,</span> + <span class="s2">"trakt-api-key"</span><span class="p">:</span> <span class="n">trakt_id</span> +<span class="p">}</span> + +<span class="n">params</span> <span class="o">=</span> <span class="p">{</span> + <span class="s2">"query"</span><span class="p">:</span> <span class="s2">""</span><span class="p">,</span> + <span class="s2">"years"</span><span class="p">:</span> <span class="s2">"1900-2021"</span><span class="p">,</span> + <span class="s2">"page"</span><span class="p">:</span> <span class="s2">"1"</span><span class="p">,</span> + <span class="s2">"extended"</span><span class="p">:</span> <span class="s2">"full"</span><span class="p">,</span> + <span class="s2">"languages"</span><span class="p">:</span> <span class="s2">"en"</span> +<span class="p">}</span> + +<span class="n">res</span> <span class="o">=</span> <span class="n">requests</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="sa">f</span><span class="s2">"</span><span class="si">{</span><span class="n">api_base</span><span class="si">}</span><span class="s2">/search/movie"</span><span class="p">,</span><span class="n">headers</span><span class="o">=</span><span class="n">headers</span><span class="p">,</span><span class="n">params</span><span class="o">=</span><span class="n">params</span><span class="p">)</span> +<span class="n">total_items</span> <span class="o">=</span> <span class="n">res</span><span class="o">.</span><span class="n">headers</span><span class="p">[</span><span class="s2">"x-pagination-item-count"</span><span class="p">]</span> +<span class="nb">print</span><span class="p">(</span><span class="sa">f</span><span class="s2">"There are </span><span class="si">{</span><span class="n">total_items</span><span class="si">}</span><span class="s2"> movies"</span><span class="p">)</span> +</code></pre></div> + +<pre><code>There are 333946 movies +</code></pre> + +<p>First, I needed to declare the database schema in (<code>database.py</code>):</p> + +<div class="codehilite"><pre><span></span><code><span class="kn">import</span> <span class="nn">sqlalchemy</span> +<span class="kn">from</span> <span class="nn">sqlalchemy</span> <span class="kn">import</span> <span class="n">create_engine</span> +<span class="kn">from</span> <span class="nn">sqlalchemy</span> <span class="kn">import</span> <span class="n">Table</span><span class="p">,</span> <span class="n">Column</span><span class="p">,</span> <span class="n">Integer</span><span class="p">,</span> <span class="n">String</span><span class="p">,</span> <span class="n">MetaData</span><span class="p">,</span> <span class="n">ForeignKey</span><span class="p">,</span> <span class="n">PickleType</span> +<span class="kn">from</span> <span class="nn">sqlalchemy</span> <span class="kn">import</span> <span class="n">insert</span> +<span class="kn">from</span> <span class="nn">sqlalchemy.orm</span> <span class="kn">import</span> <span class="n">sessionmaker</span> +<span class="kn">from</span> <span class="nn">sqlalchemy.exc</span> <span class="kn">import</span> <span class="n">IntegrityError</span> + +<span class="n">meta</span> <span class="o">=</span> <span class="n">MetaData</span><span class="p">()</span> + +<span class="n">movies_table</span> <span class="o">=</span> <span class="n">Table</span><span class="p">(</span> + <span class="s2">"movies"</span><span class="p">,</span> + <span class="n">meta</span><span class="p">,</span> + <span class="n">Column</span><span class="p">(</span><span class="s2">"trakt_id"</span><span class="p">,</span> <span class="n">Integer</span><span class="p">,</span> <span class="n">primary_key</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> <span class="n">autoincrement</span><span class="o">=</span><span class="kc">False</span><span class="p">),</span> + <span class="n">Column</span><span class="p">(</span><span class="s2">"title"</span><span class="p">,</span> <span class="n">String</span><span class="p">),</span> + <span class="n">Column</span><span class="p">(</span><span class="s2">"overview"</span><span class="p">,</span> <span class="n">String</span><span class="p">),</span> + <span class="n">Column</span><span class="p">(</span><span class="s2">"genres"</span><span class="p">,</span> <span class="n">String</span><span class="p">),</span> + <span class="n">Column</span><span class="p">(</span><span class="s2">"year"</span><span class="p">,</span> <span class="n">Integer</span><span class="p">),</span> + <span class="n">Column</span><span class="p">(</span><span class="s2">"released"</span><span class="p">,</span> <span class="n">String</span><span class="p">),</span> + <span class="n">Column</span><span class="p">(</span><span class="s2">"runtime"</span><span class="p">,</span> <span class="n">Integer</span><span class="p">),</span> + <span class="n">Column</span><span class="p">(</span><span class="s2">"country"</span><span class="p">,</span> <span class="n">String</span><span class="p">),</span> + <span class="n">Column</span><span class="p">(</span><span class="s2">"language"</span><span class="p">,</span> <span class="n">String</span><span class="p">),</span> + <span class="n">Column</span><span class="p">(</span><span class="s2">"rating"</span><span class="p">,</span> <span class="n">Integer</span><span class="p">),</span> + <span class="n">Column</span><span class="p">(</span><span class="s2">"votes"</span><span class="p">,</span> <span class="n">Integer</span><span class="p">),</span> + <span class="n">Column</span><span class="p">(</span><span class="s2">"comment_count"</span><span class="p">,</span> <span class="n">Integer</span><span class="p">),</span> + <span class="n">Column</span><span class="p">(</span><span class="s2">"tagline"</span><span class="p">,</span> <span class="n">String</span><span class="p">),</span> + <span class="n">Column</span><span class="p">(</span><span class="s2">"embeddings"</span><span class="p">,</span> <span class="n">PickleType</span><span class="p">)</span> + +<span class="p">)</span> + +<span class="c1"># Helper function to connect to the db</span> +<span class="k">def</span> <span class="nf">init_db_stuff</span><span class="p">(</span><span class="n">database_url</span><span class="p">:</span> <span class="nb">str</span><span class="p">):</span> + <span class="n">engine</span> <span class="o">=</span> <span class="n">create_engine</span><span class="p">(</span><span class="n">database_url</span><span class="p">)</span> + <span class="n">meta</span><span class="o">.</span><span class="n">create_all</span><span class="p">(</span><span class="n">engine</span><span class="p">)</span> + <span class="n">Session</span> <span class="o">=</span> <span class="n">sessionmaker</span><span class="p">(</span><span class="n">bind</span><span class="o">=</span><span class="n">engine</span><span class="p">)</span> + <span class="k">return</span> <span class="n">engine</span><span class="p">,</span> <span class="n">Session</span> +</code></pre></div> + +<p>In the end, I could have dropped the embeddings field from the table schema as I never got around to using it.</p> + +<h3>Scripting Time</h3> + +<div class="codehilite"><pre><span></span><code><span class="kn">from</span> <span class="nn">database</span> <span class="kn">import</span> <span class="o">*</span> +<span class="kn">from</span> <span class="nn">tqdm</span> <span class="kn">import</span> <span class="n">tqdm</span> +<span class="kn">import</span> <span class="nn">requests</span> +<span class="kn">import</span> <span class="nn">os</span> + +<span class="n">trakt_id</span> <span class="o">=</span> <span class="n">os</span><span class="o">.</span><span class="n">getenv</span><span class="p">(</span><span class="s2">"TRAKT_ID"</span><span class="p">)</span> + +<span class="n">max_requests</span> <span class="o">=</span> <span class="mi">5000</span> <span class="c1"># How many requests I wanted to wrap everything up in</span> +<span class="n">req_count</span> <span class="o">=</span> <span class="mi">0</span> <span class="c1"># A counter for how many requests I have made</span> + +<span class="n">years</span> <span class="o">=</span> <span class="s2">"1900-2021"</span> +<span class="n">page</span> <span class="o">=</span> <span class="mi">1</span> <span class="c1"># The initial page number for the search</span> +<span class="n">extended</span> <span class="o">=</span> <span class="s2">"full"</span> <span class="c1"># Required to get additional information </span> +<span class="n">limit</span> <span class="o">=</span> <span class="s2">"10"</span> <span class="c1"># No of entires per request -- This will be automatically picked based on max_requests</span> +<span class="n">languages</span> <span class="o">=</span> <span class="s2">"en"</span> <span class="c1"># Limit to English</span> + +<span class="n">api_base</span> <span class="o">=</span> <span class="s2">"https://api.trakt.tv"</span> +<span class="n">database_url</span> <span class="o">=</span> <span class="s2">"sqlite:///jlm.db"</span> + +<span class="n">headers</span> <span class="o">=</span> <span class="p">{</span> + <span class="s2">"Content-Type"</span><span class="p">:</span> <span class="s2">"application/json"</span><span class="p">,</span> + <span class="s2">"trakt-api-version"</span><span class="p">:</span> <span class="s2">"2"</span><span class="p">,</span> + <span class="s2">"trakt-api-key"</span><span class="p">:</span> <span class="n">trakt_id</span> +<span class="p">}</span> + +<span class="n">params</span> <span class="o">=</span> <span class="p">{</span> + <span class="s2">"query"</span><span class="p">:</span> <span class="s2">""</span><span class="p">,</span> + <span class="s2">"years"</span><span class="p">:</span> <span class="n">years</span><span class="p">,</span> + <span class="s2">"page"</span><span class="p">:</span> <span class="n">page</span><span class="p">,</span> + <span class="s2">"extended"</span><span class="p">:</span> <span class="n">extended</span><span class="p">,</span> + <span class="s2">"limit"</span><span class="p">:</span> <span class="n">limit</span><span class="p">,</span> + <span class="s2">"languages"</span><span class="p">:</span> <span class="n">languages</span> +<span class="p">}</span> + +<span class="c1"># Helper function to get desirable values from the response</span> +<span class="k">def</span> <span class="nf">create_movie_dict</span><span class="p">(</span><span class="n">movie</span><span class="p">:</span> <span class="nb">dict</span><span class="p">):</span> + <span class="n">m</span> <span class="o">=</span> <span class="n">movie</span><span class="p">[</span><span class="s2">"movie"</span><span class="p">]</span> + <span class="n">movie_dict</span> <span class="o">=</span> <span class="p">{</span> + <span class="s2">"title"</span><span class="p">:</span> <span class="n">m</span><span class="p">[</span><span class="s2">"title"</span><span class="p">],</span> + <span class="s2">"overview"</span><span class="p">:</span> <span class="n">m</span><span class="p">[</span><span class="s2">"overview"</span><span class="p">],</span> + <span class="s2">"genres"</span><span class="p">:</span> <span class="n">m</span><span class="p">[</span><span class="s2">"genres"</span><span class="p">],</span> + <span class="s2">"language"</span><span class="p">:</span> <span class="n">m</span><span class="p">[</span><span class="s2">"language"</span><span class="p">],</span> + <span class="s2">"year"</span><span class="p">:</span> <span class="nb">int</span><span class="p">(</span><span class="n">m</span><span class="p">[</span><span class="s2">"year"</span><span class="p">]),</span> + <span class="s2">"trakt_id"</span><span class="p">:</span> <span class="n">m</span><span class="p">[</span><span class="s2">"ids"</span><span class="p">][</span><span class="s2">"trakt"</span><span class="p">],</span> + <span class="s2">"released"</span><span class="p">:</span> <span class="n">m</span><span class="p">[</span><span class="s2">"released"</span><span class="p">],</span> + <span class="s2">"runtime"</span><span class="p">:</span> <span class="nb">int</span><span class="p">(</span><span class="n">m</span><span class="p">[</span><span class="s2">"runtime"</span><span class="p">]),</span> + <span class="s2">"country"</span><span class="p">:</span> <span class="n">m</span><span class="p">[</span><span class="s2">"country"</span><span class="p">],</span> + <span class="s2">"rating"</span><span class="p">:</span> <span class="nb">int</span><span class="p">(</span><span class="n">m</span><span class="p">[</span><span class="s2">"rating"</span><span class="p">]),</span> + <span class="s2">"votes"</span><span class="p">:</span> <span class="nb">int</span><span class="p">(</span><span class="n">m</span><span class="p">[</span><span class="s2">"votes"</span><span class="p">]),</span> + <span class="s2">"comment_count"</span><span class="p">:</span> <span class="nb">int</span><span class="p">(</span><span class="n">m</span><span class="p">[</span><span class="s2">"comment_count"</span><span class="p">]),</span> + <span class="s2">"tagline"</span><span class="p">:</span> <span class="n">m</span><span class="p">[</span><span class="s2">"tagline"</span><span class="p">]</span> + <span class="p">}</span> + <span class="k">return</span> <span class="n">movie_dict</span> + +<span class="c1"># Get total number of items</span> +<span class="n">params</span><span class="p">[</span><span class="s2">"limit"</span><span class="p">]</span> <span class="o">=</span> <span class="mi">1</span> +<span class="n">res</span> <span class="o">=</span> <span class="n">requests</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="sa">f</span><span class="s2">"</span><span class="si">{</span><span class="n">api_base</span><span class="si">}</span><span class="s2">/search/movie"</span><span class="p">,</span><span class="n">headers</span><span class="o">=</span><span class="n">headers</span><span class="p">,</span><span class="n">params</span><span class="o">=</span><span class="n">params</span><span class="p">)</span> +<span class="n">total_items</span> <span class="o">=</span> <span class="n">res</span><span class="o">.</span><span class="n">headers</span><span class="p">[</span><span class="s2">"x-pagination-item-count"</span><span class="p">]</span> + +<span class="n">engine</span><span class="p">,</span> <span class="n">Session</span> <span class="o">=</span> <span class="n">init_db_stuff</span><span class="p">(</span><span class="n">database_url</span><span class="p">)</span> + + +<span class="k">for</span> <span class="n">page</span> <span class="ow">in</span> <span class="n">tqdm</span><span class="p">(</span><span class="nb">range</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span><span class="n">max_requests</span><span class="o">+</span><span class="mi">1</span><span class="p">)):</span> + <span class="n">params</span><span class="p">[</span><span class="s2">"page"</span><span class="p">]</span> <span class="o">=</span> <span class="n">page</span> + <span class="n">params</span><span class="p">[</span><span class="s2">"limit"</span><span class="p">]</span> <span class="o">=</span> <span class="nb">int</span><span class="p">(</span><span class="nb">int</span><span class="p">(</span><span class="n">total_items</span><span class="p">)</span><span class="o">/</span><span class="n">max_requests</span><span class="p">)</span> + <span class="n">movies</span> <span class="o">=</span> <span class="p">[]</span> + <span class="n">res</span> <span class="o">=</span> <span class="n">requests</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="sa">f</span><span class="s2">"</span><span class="si">{</span><span class="n">api_base</span><span class="si">}</span><span class="s2">/search/movie"</span><span class="p">,</span><span class="n">headers</span><span class="o">=</span><span class="n">headers</span><span class="p">,</span><span class="n">params</span><span class="o">=</span><span class="n">params</span><span class="p">)</span> + + <span class="k">if</span> <span class="n">res</span><span class="o">.</span><span class="n">status_code</span> <span class="o">==</span> <span class="mi">500</span><span class="p">:</span> + <span class="k">break</span> + <span class="k">elif</span> <span class="n">res</span><span class="o">.</span><span class="n">status_code</span> <span class="o">==</span> <span class="mi">200</span><span class="p">:</span> + <span class="kc">None</span> + <span class="k">else</span><span class="p">:</span> + <span class="nb">print</span><span class="p">(</span><span class="sa">f</span><span class="s2">"OwO Code </span><span class="si">{</span><span class="n">res</span><span class="o">.</span><span class="n">status_code</span><span class="si">}</span><span class="s2">"</span><span class="p">)</span> + + <span class="k">for</span> <span class="n">movie</span> <span class="ow">in</span> <span class="n">res</span><span class="o">.</span><span class="n">json</span><span class="p">():</span> + <span class="n">movies</span><span class="o">.</span><span class="n">append</span><span class="p">(</span><span class="n">create_movie_dict</span><span class="p">(</span><span class="n">movie</span><span class="p">))</span> + + <span class="k">with</span> <span class="n">engine</span><span class="o">.</span><span class="n">connect</span><span class="p">()</span> <span class="k">as</span> <span class="n">conn</span><span class="p">:</span> + <span class="k">for</span> <span class="n">movie</span> <span class="ow">in</span> <span class="n">movies</span><span class="p">:</span> + <span class="k">with</span> <span class="n">conn</span><span class="o">.</span><span class="n">begin</span><span class="p">()</span> <span class="k">as</span> <span class="n">trans</span><span class="p">:</span> + <span class="n">stmt</span> <span class="o">=</span> <span class="n">insert</span><span class="p">(</span><span class="n">movies_table</span><span class="p">)</span><span class="o">.</span><span class="n">values</span><span class="p">(</span> + <span class="n">trakt_id</span><span class="o">=</span><span class="n">movie</span><span class="p">[</span><span class="s2">"trakt_id"</span><span class="p">],</span> <span class="n">title</span><span class="o">=</span><span class="n">movie</span><span class="p">[</span><span class="s2">"title"</span><span class="p">],</span> <span class="n">genres</span><span class="o">=</span><span class="s2">" "</span><span class="o">.</span><span class="n">join</span><span class="p">(</span><span class="n">movie</span><span class="p">[</span><span class="s2">"genres"</span><span class="p">]),</span> + <span class="n">language</span><span class="o">=</span><span class="n">movie</span><span class="p">[</span><span class="s2">"language"</span><span class="p">],</span> <span class="n">year</span><span class="o">=</span><span class="n">movie</span><span class="p">[</span><span class="s2">"year"</span><span class="p">],</span> <span class="n">released</span><span class="o">=</span><span class="n">movie</span><span class="p">[</span><span class="s2">"released"</span><span class="p">],</span> + <span class="n">runtime</span><span class="o">=</span><span class="n">movie</span><span class="p">[</span><span class="s2">"runtime"</span><span class="p">],</span> <span class="n">country</span><span class="o">=</span><span class="n">movie</span><span class="p">[</span><span class="s2">"country"</span><span class="p">],</span> <span class="n">overview</span><span class="o">=</span><span class="n">movie</span><span class="p">[</span><span class="s2">"overview"</span><span class="p">],</span> + <span class="n">rating</span><span class="o">=</span><span class="n">movie</span><span class="p">[</span><span class="s2">"rating"</span><span class="p">],</span> <span class="n">votes</span><span class="o">=</span><span class="n">movie</span><span class="p">[</span><span class="s2">"votes"</span><span class="p">],</span> <span class="n">comment_count</span><span class="o">=</span><span class="n">movie</span><span class="p">[</span><span class="s2">"comment_count"</span><span class="p">],</span> + <span class="n">tagline</span><span class="o">=</span><span class="n">movie</span><span class="p">[</span><span class="s2">"tagline"</span><span class="p">])</span> + <span class="k">try</span><span class="p">:</span> + <span class="n">result</span> <span class="o">=</span> <span class="n">conn</span><span class="o">.</span><span class="n">execute</span><span class="p">(</span><span class="n">stmt</span><span class="p">)</span> + <span class="n">trans</span><span class="o">.</span><span class="n">commit</span><span class="p">()</span> + <span class="k">except</span> <span class="n">IntegrityError</span><span class="p">:</span> + <span class="n">trans</span><span class="o">.</span><span class="n">rollback</span><span class="p">()</span> + <span class="n">req_count</span> <span class="o">+=</span> <span class="mi">1</span> +</code></pre></div> + +<p>(Note: I was well within the rate-limit so I did not have to slow down or implement any other measures)</p> + +<p>Running this script took me approximately 3 hours, and resulted in an SQLite database of 141.5 MB</p> + +<h2>Embeddings!</h2> + +<p>I did not want to put my poor Mac through the estimated 23 hours it would have taken to embed the sentences. I decided to use Google Colab instead.</p> + +<p>Because of the small size of the database file, I was able to just upload the file.</p> + +<p>For the encoding model, I decided to use the pretrained <code>paraphrase-multilingual-MiniLM-L12-v2</code> model for SentenceTransformers, a Python framework for SOTA sentence, text and image embeddings. I wanted to use a multilingual model as I personally consume content in various languages (natively, no dubs or subs) and some of the sources for their information do not translate to English. As of writing this post, I did not include any other database except Trakt. </p> + +<p>While deciding how I was going to process the embeddings, I came across multiple solutions:</p> + +<ul> +<li><p><a rel="noopener" target="_blank" href="https://milvus.io">Milvus</a> - An open-source vector database with similar search functionality</p></li> +<li><p><a rel="noopener" target="_blank" href="https://faiss.ai">FAISS</a> - A library for efficient similarity search</p></li> +<li><p><a rel="noopener" target="_blank" href="https://pinecone.io">Pinecone</a> - A fully managed vector database with similar search functionality</p></li> +</ul> + +<p>I did not want to waste time setting up the first two, so I decided to go with Pinecone which offers 1M 768-dim vectors for free with no credit card required (Our embeddings are 384-dim dense).</p> + +<p>Getting started with Pinecone was as easy as:</p> + +<ul> +<li><p>Signing up</p></li> +<li><p>Specifying the index name and vector dimensions along with the similarity search metric (Cosine Similarity for our use case)</p></li> +<li><p>Getting the API key</p></li> +<li><p>Installing the Python module (pinecone-client)</p></li> +</ul> + +<div class="codehilite"><pre><span></span><code><span class="kn">import</span> <span class="nn">pandas</span> <span class="k">as</span> <span class="nn">pd</span> +<span class="kn">import</span> <span class="nn">pinecone</span> +<span class="kn">from</span> <span class="nn">sentence_transformers</span> <span class="kn">import</span> <span class="n">SentenceTransformer</span> +<span class="kn">from</span> <span class="nn">tqdm</span> <span class="kn">import</span> <span class="n">tqdm</span> + +<span class="n">database_url</span> <span class="o">=</span> <span class="s2">"sqlite:///jlm.db"</span> +<span class="n">PINECONE_KEY</span> <span class="o">=</span> <span class="s2">"not-this-at-all"</span> +<span class="n">batch_size</span> <span class="o">=</span> <span class="mi">32</span> + +<span class="n">pinecone</span><span class="o">.</span><span class="n">init</span><span class="p">(</span><span class="n">api_key</span><span class="o">=</span><span class="n">PINECONE_KEY</span><span class="p">,</span> <span class="n">environment</span><span class="o">=</span><span class="s2">"us-west1-gcp"</span><span class="p">)</span> +<span class="n">index</span> <span class="o">=</span> <span class="n">pinecone</span><span class="o">.</span><span class="n">Index</span><span class="p">(</span><span class="s2">"movies"</span><span class="p">)</span> + +<span class="n">model</span> <span class="o">=</span> <span class="n">SentenceTransformer</span><span class="p">(</span><span class="s2">"paraphrase-multilingual-MiniLM-L12-v2"</span><span class="p">,</span> <span class="n">device</span><span class="o">=</span><span class="s2">"cuda"</span><span class="p">)</span> +<span class="n">engine</span><span class="p">,</span> <span class="n">Session</span> <span class="o">=</span> <span class="n">init_db_stuff</span><span class="p">(</span><span class="n">database_url</span><span class="p">)</span> + +<span class="n">df</span> <span class="o">=</span> <span class="n">pd</span><span class="o">.</span><span class="n">read_sql</span><span class="p">(</span><span class="s2">"Select * from movies"</span><span class="p">,</span> <span class="n">engine</span><span class="p">)</span> +<span class="n">df</span><span class="p">[</span><span class="s2">"combined_text"</span><span class="p">]</span> <span class="o">=</span> <span class="n">df</span><span class="p">[</span><span class="s2">"title"</span><span class="p">]</span> <span class="o">+</span> <span class="s2">": "</span> <span class="o">+</span> <span class="n">df</span><span class="p">[</span><span class="s2">"overview"</span><span class="p">]</span><span class="o">.</span><span class="n">fillna</span><span class="p">(</span><span class="s1">''</span><span class="p">)</span> <span class="o">+</span> <span class="s2">" - "</span> <span class="o">+</span> <span class="n">df</span><span class="p">[</span><span class="s2">"tagline"</span><span class="p">]</span><span class="o">.</span><span class="n">fillna</span><span class="p">(</span><span class="s1">''</span><span class="p">)</span> <span class="o">+</span> <span class="s2">" Genres:- "</span> <span class="o">+</span> <span class="n">df</span><span class="p">[</span><span class="s2">"genres"</span><span class="p">]</span><span class="o">.</span><span class="n">fillna</span><span class="p">(</span><span class="s1">''</span><span class="p">)</span> + +<span class="c1"># Creating the embedding and inserting it into the database</span> +<span class="k">for</span> <span class="n">x</span> <span class="ow">in</span> <span class="n">tqdm</span><span class="p">(</span><span class="nb">range</span><span class="p">(</span><span class="mi">0</span><span class="p">,</span><span class="nb">len</span><span class="p">(</span><span class="n">df</span><span class="p">),</span><span class="n">batch_size</span><span class="p">)):</span> + <span class="n">to_send</span> <span class="o">=</span> <span class="p">[]</span> + <span class="n">trakt_ids</span> <span class="o">=</span> <span class="n">df</span><span class="p">[</span><span class="s2">"trakt_id"</span><span class="p">][</span><span class="n">x</span><span class="p">:</span><span class="n">x</span><span class="o">+</span><span class="n">batch_size</span><span class="p">]</span><span class="o">.</span><span class="n">tolist</span><span class="p">()</span> + <span class="n">sentences</span> <span class="o">=</span> <span class="n">df</span><span class="p">[</span><span class="s2">"combined_text"</span><span class="p">][</span><span class="n">x</span><span class="p">:</span><span class="n">x</span><span class="o">+</span><span class="n">batch_size</span><span class="p">]</span><span class="o">.</span><span class="n">tolist</span><span class="p">()</span> + <span class="n">embeddings</span> <span class="o">=</span> <span class="n">model</span><span class="o">.</span><span class="n">encode</span><span class="p">(</span><span class="n">sentences</span><span class="p">)</span> + <span class="k">for</span> <span class="n">idx</span><span class="p">,</span> <span class="n">value</span> <span class="ow">in</span> <span class="nb">enumerate</span><span class="p">(</span><span class="n">trakt_ids</span><span class="p">):</span> + <span class="n">to_send</span><span class="o">.</span><span class="n">append</span><span class="p">(</span> + <span class="p">(</span> + <span class="nb">str</span><span class="p">(</span><span class="n">value</span><span class="p">),</span> <span class="n">embeddings</span><span class="p">[</span><span class="n">idx</span><span class="p">]</span><span class="o">.</span><span class="n">tolist</span><span class="p">()</span> + <span class="p">))</span> + <span class="n">index</span><span class="o">.</span><span class="n">upsert</span><span class="p">(</span><span class="n">to_send</span><span class="p">)</span> +</code></pre></div> + +<p>That's it!</p> + +<h2>Interacting with Vectors</h2> + +<p>We use the <code>trakt_id</code> for the movie as the ID for the vectors and upsert it into the index. </p> + +<p>To find similar items, we will first have to map the name of the movie to its trakt_id, get the embeddings we have for that id and then perform a similarity search. It is possible that this additional step of mapping could be avoided by storing information as metadata in the index.</p> + +<div class="codehilite"><pre><span></span><code><span class="k">def</span> <span class="nf">get_trakt_id</span><span class="p">(</span><span class="n">df</span><span class="p">,</span> <span class="n">title</span><span class="p">:</span> <span class="nb">str</span><span class="p">):</span> + <span class="n">rec</span> <span class="o">=</span> <span class="n">df</span><span class="p">[</span><span class="n">df</span><span class="p">[</span><span class="s2">"title"</span><span class="p">]</span><span class="o">.</span><span class="n">str</span><span class="o">.</span><span class="n">lower</span><span class="p">()</span><span class="o">==</span><span class="n">movie_name</span><span class="o">.</span><span class="n">lower</span><span class="p">()]</span> + <span class="k">if</span> <span class="nb">len</span><span class="p">(</span><span class="n">rec</span><span class="o">.</span><span class="n">trakt_id</span><span class="o">.</span><span class="n">values</span><span class="o">.</span><span class="n">tolist</span><span class="p">())</span> <span class="o">></span> <span class="mi">1</span><span class="p">:</span> + <span class="nb">print</span><span class="p">(</span><span class="sa">f</span><span class="s2">"multiple values found... </span><span class="si">{</span><span class="nb">len</span><span class="p">(</span><span class="n">rec</span><span class="o">.</span><span class="n">trakt_id</span><span class="o">.</span><span class="n">values</span><span class="p">)</span><span class="si">}</span><span class="s2">"</span><span class="p">)</span> + <span class="k">for</span> <span class="n">x</span> <span class="ow">in</span> <span class="nb">range</span><span class="p">(</span><span class="nb">len</span><span class="p">(</span><span class="n">rec</span><span class="p">)):</span> + <span class="nb">print</span><span class="p">(</span><span class="sa">f</span><span class="s2">"[</span><span class="si">{</span><span class="n">x</span><span class="si">}</span><span class="s2">] </span><span class="si">{</span><span class="n">rec</span><span class="p">[</span><span class="s1">'title'</span><span class="p">]</span><span class="o">.</span><span class="n">tolist</span><span class="p">()[</span><span class="n">x</span><span class="p">]</span><span class="si">}</span><span class="s2"> (</span><span class="si">{</span><span class="n">rec</span><span class="p">[</span><span class="s1">'year'</span><span class="p">]</span><span class="o">.</span><span class="n">tolist</span><span class="p">()[</span><span class="n">x</span><span class="p">]</span><span class="si">}</span><span class="s2">) - </span><span class="si">{</span><span class="n">rec</span><span class="p">[</span><span class="s1">'overview'</span><span class="p">]</span><span class="o">.</span><span class="n">tolist</span><span class="p">()</span><span class="si">}</span><span class="s2">"</span><span class="p">)</span> + <span class="nb">print</span><span class="p">(</span><span class="s2">"==="</span><span class="p">)</span> + <span class="n">z</span> <span class="o">=</span> <span class="nb">int</span><span class="p">(</span><span class="nb">input</span><span class="p">(</span><span class="s2">"Choose No: "</span><span class="p">))</span> + <span class="k">return</span> <span class="n">rec</span><span class="o">.</span><span class="n">trakt_id</span><span class="o">.</span><span class="n">values</span><span class="p">[</span><span class="n">z</span><span class="p">]</span> + <span class="k">return</span> <span class="n">rec</span><span class="o">.</span><span class="n">trakt_id</span><span class="o">.</span><span class="n">values</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span> + +<span class="k">def</span> <span class="nf">get_vector_value</span><span class="p">(</span><span class="n">trakt_id</span><span class="p">:</span> <span class="nb">int</span><span class="p">):</span> + <span class="n">fetch_response</span> <span class="o">=</span> <span class="n">index</span><span class="o">.</span><span class="n">fetch</span><span class="p">(</span><span class="n">ids</span><span class="o">=</span><span class="p">[</span><span class="nb">str</span><span class="p">(</span><span class="n">trakt_id</span><span class="p">)])</span> + <span class="k">return</span> <span class="n">fetch_response</span><span class="p">[</span><span class="s2">"vectors"</span><span class="p">][</span><span class="nb">str</span><span class="p">(</span><span class="n">trakt_id</span><span class="p">)][</span><span class="s2">"values"</span><span class="p">]</span> + +<span class="k">def</span> <span class="nf">query_vectors</span><span class="p">(</span><span class="n">vector</span><span class="p">:</span> <span class="nb">list</span><span class="p">,</span> <span class="n">top_k</span><span class="p">:</span> <span class="nb">int</span> <span class="o">=</span> <span class="mi">20</span><span class="p">,</span> <span class="n">include_values</span><span class="p">:</span> <span class="nb">bool</span> <span class="o">=</span> <span class="kc">False</span><span class="p">,</span> <span class="n">include_metada</span><span class="p">:</span> <span class="nb">bool</span> <span class="o">=</span> <span class="kc">True</span><span class="p">):</span> + <span class="n">query_response</span> <span class="o">=</span> <span class="n">index</span><span class="o">.</span><span class="n">query</span><span class="p">(</span> + <span class="n">queries</span><span class="o">=</span><span class="p">[</span> + <span class="p">(</span><span class="n">vector</span><span class="p">),</span> + <span class="p">],</span> + <span class="n">top_k</span><span class="o">=</span><span class="n">top_k</span><span class="p">,</span> + <span class="n">include_values</span><span class="o">=</span><span class="n">include_values</span><span class="p">,</span> + <span class="n">include_metadata</span><span class="o">=</span><span class="n">include_metada</span> + <span class="p">)</span> + <span class="k">return</span> <span class="n">query_response</span> + +<span class="k">def</span> <span class="nf">query2ids</span><span class="p">(</span><span class="n">query_response</span><span class="p">):</span> + <span class="n">trakt_ids</span> <span class="o">=</span> <span class="p">[]</span> + <span class="k">for</span> <span class="n">match</span> <span class="ow">in</span> <span class="n">query_response</span><span class="p">[</span><span class="s2">"results"</span><span class="p">][</span><span class="mi">0</span><span class="p">][</span><span class="s2">"matches"</span><span class="p">]:</span> + <span class="n">trakt_ids</span><span class="o">.</span><span class="n">append</span><span class="p">(</span><span class="nb">int</span><span class="p">(</span><span class="n">match</span><span class="p">[</span><span class="s2">"id"</span><span class="p">]))</span> + <span class="k">return</span> <span class="n">trakt_ids</span> + +<span class="k">def</span> <span class="nf">get_deets_by_trakt_id</span><span class="p">(</span><span class="n">df</span><span class="p">,</span> <span class="n">trakt_id</span><span class="p">:</span> <span class="nb">int</span><span class="p">):</span> + <span class="n">df</span> <span class="o">=</span> <span class="n">df</span><span class="p">[</span><span class="n">df</span><span class="p">[</span><span class="s2">"trakt_id"</span><span class="p">]</span><span class="o">==</span><span class="n">trakt_id</span><span class="p">]</span> + <span class="k">return</span> <span class="p">{</span> + <span class="s2">"title"</span><span class="p">:</span> <span class="n">df</span><span class="o">.</span><span class="n">title</span><span class="o">.</span><span class="n">values</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span> + <span class="s2">"overview"</span><span class="p">:</span> <span class="n">df</span><span class="o">.</span><span class="n">overview</span><span class="o">.</span><span class="n">values</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span> + <span class="s2">"runtime"</span><span class="p">:</span> <span class="n">df</span><span class="o">.</span><span class="n">runtime</span><span class="o">.</span><span class="n">values</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span> + <span class="s2">"year"</span><span class="p">:</span> <span class="n">df</span><span class="o">.</span><span class="n">year</span><span class="o">.</span><span class="n">values</span><span class="p">[</span><span class="mi">0</span><span class="p">]</span> + <span class="p">}</span> +</code></pre></div> + +<h3>Testing it Out</h3> + +<div class="codehilite"><pre><span></span><code><span class="n">movie_name</span> <span class="o">=</span> <span class="s2">"Now You See Me"</span> + +<span class="n">movie_trakt_id</span> <span class="o">=</span> <span class="n">get_trakt_id</span><span class="p">(</span><span class="n">df</span><span class="p">,</span> <span class="n">movie_name</span><span class="p">)</span> +<span class="nb">print</span><span class="p">(</span><span class="n">movie_trakt_id</span><span class="p">)</span> +<span class="n">movie_vector</span> <span class="o">=</span> <span class="n">get_vector_value</span><span class="p">(</span><span class="n">movie_trakt_id</span><span class="p">)</span> +<span class="n">movie_queries</span> <span class="o">=</span> <span class="n">query_vectors</span><span class="p">(</span><span class="n">movie_vector</span><span class="p">)</span> +<span class="n">movie_ids</span> <span class="o">=</span> <span class="n">query2ids</span><span class="p">(</span><span class="n">movie_queries</span><span class="p">)</span> +<span class="nb">print</span><span class="p">(</span><span class="n">movie_ids</span><span class="p">)</span> + +<span class="k">for</span> <span class="n">trakt_id</span> <span class="ow">in</span> <span class="n">movie_ids</span><span class="p">:</span> + <span class="n">deets</span> <span class="o">=</span> <span class="n">get_deets_by_trakt_id</span><span class="p">(</span><span class="n">df</span><span class="p">,</span> <span class="n">trakt_id</span><span class="p">)</span> + <span class="nb">print</span><span class="p">(</span><span class="sa">f</span><span class="s2">"</span><span class="si">{</span><span class="n">deets</span><span class="p">[</span><span class="s1">'title'</span><span class="p">]</span><span class="si">}</span><span class="s2"> (</span><span class="si">{</span><span class="n">deets</span><span class="p">[</span><span class="s1">'year'</span><span class="p">]</span><span class="si">}</span><span class="s2">): </span><span class="si">{</span><span class="n">deets</span><span class="p">[</span><span class="s1">'overview'</span><span class="p">]</span><span class="si">}</span><span class="s2">"</span><span class="p">)</span> +</code></pre></div> + +<p>Output:</p> + +<pre><code>[55786, 18374, 299592, 662622, 6054, 227458, 139687, 303950, 70000, 129307, 70823, 5766, 23950, 137696, 655723, 32842, 413269, 145994, 197990, 373832] +Now You See Me (2013): An FBI agent and an Interpol detective track a team of illusionists who pull off bank heists during their performances and reward their audiences with the money. +Trapped (1949): U.S. Treasury Department agents go after a ring of counterfeiters. +Brute Sanity (2018): An FBI-trained neuropsychologist teams up with a thief to find a reality-altering device while her insane ex-boss unleashes bizarre traps to stop her. +The Chase (2017): Some FBI agents hunt down a criminal +Surveillance (2008): An FBI agent tracks a serial killer with the help of three of his would-be victims - all of whom have wildly different stories to tell. +Marauders (2016): An untraceable group of elite bank robbers is chased by a suicidal FBI agent who uncovers a deeper purpose behind the robbery-homicides. +Miracles for Sale (1939): A maker of illusions for magicians protects an ingenue likely to be murdered. +Deceptors (2005): A Ghostbusters knock-off where a group of con-artists create bogus monsters to scare up some cash. They run for their lives when real spooks attack. +The Outfit (1993): A renegade FBI agent sparks an explosive mob war between gangster crime lords Legs Diamond and Dutch Schultz. +Bank Alarm (1937): A federal agent learns the gangsters he's been investigating have kidnapped his sister. +The Courier (2012): A shady FBI agent recruits a courier to deliver a mysterious package to a vengeful master criminal who has recently resurfaced with a diabolical plan. +After the Sunset (2004): An FBI agent is suspicious of two master thieves, quietly enjoying their retirement near what may - or may not - be the biggest score of their careers. +Down Three Dark Streets (1954): An FBI Agent takes on the three unrelated cases of a dead agent to track down his killer. +The Executioner (1970): A British intelligence agent must track down a fellow spy suspected of being a double agent. +Ace of Cactus Range (1924): A Secret Service agent goes undercover to unmask the leader of a gang of diamond thieves. +Firepower (1979): A mercenary is hired by the FBI to track down a powerful recluse criminal, a woman is also trying to track him down for her own personal vendetta. +Heroes & Villains (2018): an FBI agent chases a thug to great tunes +Federal Fugitives (1941): A government agent goes undercover in order to apprehend a saboteur who caused a plane crash. +Hell on Earth (2012): An FBI Agent on the trail of a group of drug traffickers learns that their corruption runs deeper than she ever imagined, and finds herself in a supernatural - and deadly - situation. +Spies (2015): A secret agent must perform a heist without time on his side +</code></pre> + +<p>For now, I am happy with the recommendations.</p> + +<h2>Simple UI</h2> + +<p>The code for the flask app can be found on GitHub: <a rel="noopener" target="_blank" href="https://github.com/navanchauhan/FlixRec">navanchauhan/FlixRec</a> or on my <a rel="noopener" target="_blank" href="https://pi4.navan.dev/gitea/navan/FlixRec">Gitea instance</a></p> + +<p>I quickly whipped up a simple Flask App to deal with problems of multiple movies sharing the title, and typos in the search query.</p> + +<h3>Home Page</h3> + +<p><img src="/assets/flixrec/home.png" alt="Home Page" /></p> + +<h3>Handling Multiple Movies with Same Title</h3> + +<p><img src="/assets/flixrec/multiple.png" alt="Multiple Movies with Same Title" /></p> + +<h3>Results Page</h3> + +<p><img src="/assets/flixrec/results.png" alt="Results Page" /></p> + +<p>Includes additional filter options</p> + +<p><img src="/assets/flixrec/filter.png" alt="Advance Filtering Options" /></p> + +<p>Test it out at <a rel="noopener" target="_blank" href="https://flixrec.navan.dev">https://flixrec.navan.dev</a></p> + +<h2>Current Limittations</h2> + +<ul> +<li>Does not work well with popular franchises</li> +<li>No Genre Filter</li> +</ul> + +<h2>Future Addons</h2> + +<ul> +<li>Include Cast Data +<ul> +<li>e.g. If it sees a movie with Tom Hanks and Meg Ryan, then it will boost similar movies including them</li> +<li>e.g. If it sees the movie has been directed my McG, then it will boost similar movies directed by them</li> +</ul></li> +<li>REST API</li> +<li>TV Shows</li> +<li>Multilingual database</li> +<li>Filter based on popularity: The data already exists in the indexed database</li> +</ul> + +</main> + + +<script src="assets/manup.min.js"></script> +<script src="/pwabuilder-sw-register.js"></script> +</body> +</html>
\ No newline at end of file diff --git a/docs/posts/index.html b/docs/posts/index.html index 223e6d3..824554c 100644 --- a/docs/posts/index.html +++ b/docs/posts/index.html @@ -50,6 +50,23 @@ <ul> + <li><a href="/posts/2022-05-21-Similar-Movies-Recommender.html">Building a Simple Similar Movies Recommender System</a></li> + <ul> + <li>Building a Content Based Similar Movies Recommender System</li> + <li>Published On: 2022-05-21 17:56</li> + <li>Tags: + + Python, + + Transformers, + + Movies, + + Recommender-System, + + </ul> + + <li><a href="/posts/2021-06-27-Crude-ML-AI-Powered-Chatbot-Swift.html">Making a Crude ML Powered Chatbot in Swift using CoreML</a></li> <ul> <li>Writing a simple Machine-Learning powered Chatbot (or, daresay virtual personal assistant ) in Swift using CoreML.</li> |