more metric citations

lenskit · Feb 12, 2022 · d88b41b · d88b41b
1 parent 7b12ec3
commit d88b41b
Show file tree

Hide file tree

Showing 2 changed files with 20 additions and 7 deletions.
diff --git a/docs/lenskit.bib b/docs/lenskit.bib
@@ -146,6 +146,17 @@ @ARTICLE{movielens
   doi      = "10.1145/2827872"
 }
 
+@INPROCEEDINGS{Kantor1997-lm,
+  title           = "Report on the {TREC-5} Confusion Track",
+  booktitle       = "The Fifth Text {REtrieval} Conference ({TREC-5})",
+  author          = "Kantor, Paul B and Voorhees, Ellen",
+  institution     = "NIST",
+  month           =  oct,
+  year            =  1997,
+  url             = "http://trec.nist.gov/pubs/trec5/t5_proceedings.html",
+  location        = "Gaithersburg, Maryland"
+}
+
 @INPROCEEDINGS{Hu2008-li,
   title     = "Collaborative Filtering for Implicit Feedback Datasets",
   booktitle = "2008 Eighth {IEEE} International Conference on Data Mining",

diff --git a/lenskit/metrics/topn.py b/lenskit/metrics/topn.py
@@ -105,11 +105,13 @@ def _bulk_recall(recs, truth, k=None):
 
 def recip_rank(recs, truth, k=None):
     """
-    Compute the reciprocal rank of the first relevant item in a list of recommendations.
-    Let :math:`\\kappa` denote the 1-based rank of the first relevant item in :math:`L`,
-    with :math:`\\kappa=\\infty` if none of the first :math:`k` items in :math:`L` are relevant;
-    then the reciprocal rank is :math:`1 / \\kappa`. If no elements are relevant, the reciprocal
-    rank is therefore 0.
+    Compute the reciprocal rank :cite:p:`Kantor1997-lm` of the first relevant
+    item in a list of recommendations. Let :math:`\\kappa` denote the 1-based
+    rank of the first relevant item in :math:`L`, with :math:`\\kappa=\\infty`
+    if none of the first :math:`k` items in :math:`L` are relevant; then the
+    reciprocal rank is :math:`1 / \\kappa`. If no elements are relevant, the
+    reciprocal rank is therefore 0.  :cite:t:`Deshpande2004-ht` call this the
+    “reciprocal hit rate”.
 
     This metric has a bulk equivalent.
     """
@@ -178,7 +180,7 @@ def _fixed_dcg(n, discount=np.log2):
 
 def dcg(recs, truth, discount=np.log2):
     """
-    Compute the **unnormalized** discounted cumulative gain.
+    Compute the **unnormalized** discounted cumulative gain :cite:p:`Jarvelin2002-xf`.
 
     Discounted cumultative gain is computed as:
 
@@ -213,7 +215,7 @@ def dcg(recs, truth, discount=np.log2):
 
 def ndcg(recs, truth, discount=np.log2, k=None):
     """
-    Compute the normalized discounted cumulative gain.
+    Compute the normalized discounted cumulative gain :cite:p:`Jarvelin2002-xf`.
 
     Discounted cumultative gain is computed as: