OCW completeness penalty in search

frontends/mit-learn/src/page-components/SearchDisplay/SearchDisplay.tsx

@@ -550,6 +550,9 @@ const SearchDisplay: React.FC<SearchDisplayProps> = ({
       search_mode: searchParams.get("search_mode"),
       slop: searchParams.get("slop"),
       min_score: searchParams.get("min_score"),
+      max_incompleteness_penalty: searchParams.get(
+        "max_incompleteness_penalty",
+      ),
       ...requestParams,
       aggregations: (facetNames || []).concat([
         "resource_category",
@@ -703,6 +706,28 @@ const SearchDisplay: React.FC<SearchDisplayProps> = ({
               Minimum relevance score for a search result to be displayed. Only
               affects results if there is a search term.
             </ExplanationContainer>
+            <AdminTitleContainer>
+              Maximum Incompleteness Penalty
+            </AdminTitleContainer>
+            <SliderInput
+              currentValue={
+                searchParams.get("max_incompleteness_penalty")
+                  ? Number(searchParams.get("max_incompleteness_penalty"))
+                  : 0
+              }
+              setSearchParams={setSearchParams}
+              urlParam="max_incompleteness_penalty"
+              min={0}
+              max={100}
+              step={1}
+            />
+            <ExplanationContainer>
+              Maximum score penalty for incomplete OCW courses in percent. An
+              OCW course with completeness = 0 will have this score penalty.
+              Partially complete courses have a linear penalty proportional to
+              the degree of incompleteness. Only affects results if there is a
+              search term.
+            </ExplanationContainer>
           </div>
         ) : null}
       </div>

frontends/mit-learn/src/pages/SearchPage/SearchPage.test.tsx

@@ -315,6 +315,7 @@ describe("SearchPage", () => {
     await waitFor(() => {
       screen.getByTestId("yearly_decay_percent-slider")
       screen.getByTestId("min_score-slider")
+      screen.getByTestId("max_incompleteness_penalty-slider")
     })
   })
 })

learning_resources_search/api.py

@@ -544,32 +544,48 @@ def add_text_query_to_search(search, text, search_params, query_type_query):
 
     yearly_decay_percent = search_params.get("yearly_decay_percent")
     min_score = search_params.get("min_score")
+    max_incompleteness_penalty = (
+        search_params.get("max_incompleteness_penalty", 0) / 100
+    )
 
-    if (yearly_decay_percent and yearly_decay_percent > 0) or (
-        min_score and min_score > 0
-    ):
+    if yearly_decay_percent or min_score or max_incompleteness_penalty:
         script_query = {
             "script_score": {
                 "query": {"bool": {"must": [text_query], "filter": query_type_query}}
             }
         }
 
-        if yearly_decay_percent and yearly_decay_percent > 0:
-            script_query["script_score"]["script"] = {
-                "source": (
-                    "doc['resource_age_date'].size() == 0 ? _score : "
-                    "_score * decayDateLinear(params.origin, params.scale, "
-                    "params.offset, params.decay, doc['resource_age_date'].value)"
-                ),
-                "params": {
-                    "origin": datetime.now(tz=UTC).strftime("%Y-%m-%dT%H:%M:%S.%fZ"),
-                    "offset": "0",
-                    "scale": "354d",
-                    "decay": 1 - (yearly_decay_percent / 100),
-                },
-            }
+        completeness_term = (
+            "(doc['completeness'].value * params.max_incompleteness_penalty + "
+            "(1-params.max_incompleteness_penalty))"
+        )
+
+        staleness_term = (
+            "(doc['resource_age_date'].size() == 0 ? 1 : "
+            "decayDateLinear(params.origin, params.scale, params.offset, params.decay, "
+            "doc['resource_age_date'].value))"
+        )
+
+        source = "_score"
+        params = {}
+
+        if max_incompleteness_penalty:
+            source = f"{source} * {completeness_term}"
+            params["max_incompleteness_penalty"] = max_incompleteness_penalty
+
+        if yearly_decay_percent:
+            source = f"{source} * {staleness_term}"
+            params["decay"] = 1 - (yearly_decay_percent / 100)
+            params["offset"] = "0"
+            params["scale"] = "365d"
+            params["origin"] = datetime.now(tz=UTC).strftime("%Y-%m-%dT%H:%M:%S.%fZ")
+
+        script_query["script_score"]["script"] = {
+            "source": source,
+            "params": params,
+        }
 
-        if min_score and min_score > 0:
+        if min_score:
             script_query["script_score"]["min_score"] = min_score
 
         search = search.query(script_query)

learning_resources_search/api_test.py

@@ -1440,11 +1440,55 @@ def test_execute_learn_search_for_learning_resource_query(opensearch):
 
 
 @freeze_time("2024-07-20")
-def test_execute_learn_search_with_yearly_decay_percent(mocker, opensearch):
+@pytest.mark.parametrize(
+    ("yearly_decay_percent", "max_incompleteness_penalty"),
+    [
+        (5, 25),
+        (0, 25),
+        (5, 0),
+    ],
+)
+def test_execute_learn_search_with_script_score(
+    mocker, opensearch, yearly_decay_percent, max_incompleteness_penalty
+):
     opensearch.conn.search.return_value = {
         "hits": {"total": {"value": 10, "relation": "eq"}}
     }
 
+    if yearly_decay_percent > 0 and max_incompleteness_penalty > 0:
+        source = (
+            "_score * (doc['completeness'].value * params.max_incompleteness_penalty + "
+            "(1-params.max_incompleteness_penalty)) * (doc['resource_age_date'].size() == 0 ? "
+            "1 : decayDateLinear(params.origin, params.scale, params.offset, params.decay, "
+            "doc['resource_age_date'].value))"
+        )
+        params = {
+            "origin": "2024-07-20T00:00:00.000000Z",
+            "offset": "0",
+            "scale": "365d",
+            "decay": 0.95,
+            "max_incompleteness_penalty": 0.25,
+        }
+    elif yearly_decay_percent > 0:
+        source = (
+            "_score * (doc['resource_age_date'].size() == 0 ? "
+            "1 : decayDateLinear(params.origin, params.scale, params.offset, params.decay, "
+            "doc['resource_age_date'].value))"
+        )
+
+        params = {
+            "origin": "2024-07-20T00:00:00.000000Z",
+            "offset": "0",
+            "scale": "365d",
+            "decay": 0.95,
+        }
+    else:
+        source = (
+            "_score * (doc['completeness'].value * params.max_incompleteness_penalty +"
+            " (1-params.max_incompleteness_penalty))"
+        )
+        params = {"max_incompleteness_penalty": 0.25}
+
     search_params = {
         "aggregations": ["offered_by"],
         "q": "math",
@@ -1454,7 +1498,8 @@ def test_execute_learn_search_with_yearly_decay_percent(mocker, opensearch):
         "offset": 1,
         "sortby": "-readable_id",
         "endpoint": LEARNING_RESOURCE,
-        "yearly_decay_percent": 5,
+        "yearly_decay_percent": yearly_decay_percent,
+        "max_incompleteness_penalty": max_incompleteness_penalty,
     }
 
     query = {
@@ -1700,13 +1745,8 @@ def test_execute_learn_search_with_yearly_decay_percent(mocker, opensearch):
                     }
                 },
                 "script": {
-                    "source": "doc['resource_age_date'].size() == 0 ? _score : _score * decayDateLinear(params.origin, params.scale, params.offset, params.decay, doc['resource_age_date'].value)",
-                    "params": {
-                        "origin": "2024-07-20T00:00:00.000000Z",
-                        "offset": "0",
-                        "scale": "354d",
-                        "decay": 0.95,
-                    },
+                    "source": source,
+                    "params": params,
                 },
             }
         },
@@ -2115,6 +2155,7 @@ def test_execute_learn_search_with_min_score(mocker, opensearch):
                         "filter": [{"exists": {"field": "resource_type"}}],
                     }
                 },
+                "script": {"params": {}, "source": "_score"},
                 "min_score": 5,
             }
         },

learning_resources_search/constants.py

@@ -252,6 +252,7 @@ class FilterConfig:
     "next_start_date": {"type": "date"},
     "resource_age_date": {"type": "date"},
     "featured_rank": {"type": "float"},
+    "completeness": {"type": "float"},
 }
 
 

learning_resources_search/serializers.py

@@ -430,6 +430,20 @@ class LearningResourcesSearchRequestSerializer(SearchRequestSerializer):
             "Minimum score value a text query result needs to have to be displayed"
         ),
     )
+    max_incompleteness_penalty = serializers.FloatField(
+        max_value=100,
+        min_value=0,
+        required=False,
+        allow_null=True,
+        default=0,
+        help_text=(
+            "Maximum score penalty for incomplete OCW courses in percent. "
+            "An OCW course with completeness = 0 will have this score penalty. "
+            "Partially complete courses have a linear penalty proportional to "
+            "the degree of incompleteness. Only affects results if there is a "
+            "search term."
+        ),
+    )
 
 
 class ContentFileSearchRequestSerializer(SearchRequestSerializer):

learning_resources_search/serializers_test.py

@@ -878,6 +878,7 @@ def test_learning_resources_search_request_serializer():
         "search_mode": "phrase",
         "slop": 2,
         "min_score": 0,
+        "max_incompleteness_penalty": 25,
     }
 
     cleaned = {
@@ -903,6 +904,7 @@ def test_learning_resources_search_request_serializer():
         "search_mode": "phrase",
         "slop": 2,
         "min_score": 0,
+        "max_incompleteness_penalty": 25,
     }
 
     serialized = LearningResourcesSearchRequestSerializer(data=data)

openapi/specs/v1.yaml

@@ -2450,6 +2450,19 @@ paths:
         schema:
           type: integer
         description: Number of results to return per page
+      - in: query
+        name: max_incompleteness_penalty
+        schema:
+          type: number
+          format: double
+          maximum: 100
+          minimum: 0
+          nullable: true
+          default: 0.0
+        description: Maximum score penalty for incomplete OCW courses in percent.
+          An OCW course with completeness = 0 will have this score penalty. Partially
+          complete courses have a linear penalty proportional to the degree of incompleteness.
+          Only affects results if there is a search term.
       - in: query
         name: min_score
         schema:
@@ -2918,6 +2931,19 @@ paths:
         schema:
           type: integer
         description: Number of results to return per page
+      - in: query
+        name: max_incompleteness_penalty
+        schema:
+          type: number
+          format: double
+          maximum: 100
+          minimum: 0
+          nullable: true
+          default: 0.0
+        description: Maximum score penalty for incomplete OCW courses in percent.
+          An OCW course with completeness = 0 will have this score penalty. Partially
+          complete courses have a linear penalty proportional to the degree of incompleteness.
+          Only affects results if there is a search term.
       - in: query
         name: min_score
         schema:
@@ -3411,6 +3437,19 @@ paths:
         schema:
           type: integer
         description: Number of results to return per page
+      - in: query
+        name: max_incompleteness_penalty
+        schema:
+          type: number
+          format: double
+          maximum: 100
+          minimum: 0
+          nullable: true
+          default: 0.0
+        description: Maximum score penalty for incomplete OCW courses in percent.
+          An OCW course with completeness = 0 will have this score penalty. Partially
+          complete courses have a linear penalty proportional to the degree of incompleteness.
+          Only affects results if there is a search term.
       - in: query
         name: min_score
         schema:
@@ -3895,6 +3934,19 @@ paths:
         schema:
           type: integer
         description: Number of results to return per page
+      - in: query
+        name: max_incompleteness_penalty
+        schema:
+          type: number
+          format: double
+          maximum: 100
+          minimum: 0
+          nullable: true
+          default: 0.0
+        description: Maximum score penalty for incomplete OCW courses in percent.
+          An OCW course with completeness = 0 will have this score penalty. Partially
+          complete courses have a linear penalty proportional to the degree of incompleteness.
+          Only affects results if there is a search term.
       - in: query
         name: min_score
         schema:
@@ -9913,6 +9965,17 @@ components:
           default: 0.0
           description: Minimum score value a text query result needs to have to be
             displayed
+        max_incompleteness_penalty:
+          type: number
+          format: double
+          maximum: 100
+          minimum: 0
+          nullable: true
+          default: 0.0
+          description: Maximum score penalty for incomplete OCW courses in percent.
+            An OCW course with completeness = 0 will have this score penalty. Partially
+            complete courses have a linear penalty proportional to the degree of incompleteness.
+            Only affects results if there is a search term.
         source_type:
           allOf:
           - $ref: '#/components/schemas/SourceTypeEnum'