From 4dbb8769dfc33fb46003a53f7136d4c9e8f0b8e9 Mon Sep 17 00:00:00 2001
From: Doug Turnbull <softwaredoug@gmail.com>
Date: Sat, 20 Jul 2024 07:53:22 -0400
Subject: [PATCH] More BM25 similarity tests

---
 test/test_similarity.py | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)

diff --git a/test/test_similarity.py b/test/test_similarity.py
index cb0c72e..60fae98 100644
--- a/test/test_similarity.py
+++ b/test/test_similarity.py
@@ -30,6 +30,22 @@ def arr(x):
         "num_docs": 8514,
         "expected": 3.8199246
     },
+    "rambo_tmdb": {
+        "term_freqs": 2,  # freq, occurrences of term within document
+        "doc_freqs": 7,   # n, number of documents containing term
+        "doc_lens": 44,   # "dl, length of field (approximate)
+        "avg_doc_len": 50.580456,  # avgdl, average length of field
+        "num_docs": 8514,    # N, total number of documents with field
+        "expected": 4.5636616
+    },
+    "the_tmdb": {
+        "term_freqs": 25,  # freq, occurrences of term within document
+        "doc_freqs": 7823,   # n, number of documents containing term
+        "doc_lens": 152,   # "dl, length of field (approximate)
+        "avg_doc_len": 119.18542,  # avgdl, average length of field
+        "num_docs": 8516,    # N, total number of documents with field
+        "expected": 0.08028283
+    }
 }