themefinder 0.7.0__tar.gz → 0.7.1__tar.gz

{themefinder-0.7.0 → themefinder-0.7.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: themefinder
-Version: 0.7.0
+Version: 0.7.1
 Summary: A topic modelling Python package designed for analysing one-to-many question-answer data eg free-text survey responses.
 License: MIT
 Author: i.AI
@@ -17,7 +17,7 @@ Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
 Classifier: Topic :: Text Processing :: Linguistic
 Requires-Dist: boto3 (>=1.29,<2.0)
 Requires-Dist: langchain
-Requires-Dist: langchain-openai (==0.1.17)
+Requires-Dist: langchain-openai
 Requires-Dist: langfuse (==2.29.1)
 Requires-Dist: openpyxl (>=3.1.5,<4.0.0)
 Requires-Dist: pandas (>=2.2.2,<3.0.0)

{themefinder-0.7.0 → themefinder-0.7.1}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "themefinder"
-version = "0.7.0"
+version = "0.7.1"
 description = "A topic modelling Python package designed for analysing one-to-many question-answer data eg free-text survey responses."
 authors = ["i.AI <packages@cabinetoffice.gov.uk>"]
 packages = [{include = "themefinder", from = "src"}]
@@ -19,7 +19,7 @@ classifiers = [
 [tool.poetry.dependencies]
 python = ">=3.10,<3.13"
 langchain = "*"
-langchain-openai = "0.1.17"
+langchain-openai = "*"
 pandas = "^2.2.2"
 python-dotenv = "^1.0.1"
 langfuse = "2.29.1"

themefinder-0.7.1/src/themefinder/prompts/detail_detection.txt

@@ -0,0 +1,31 @@
+{system_prompt}
+
+You will receive a list of RESPONSES, each containing a response_id and a response.
+Your job is to analyze each response to the QUESTION below and decide if a response contains rich evidence.
+You MUST include every response ID in the output.
+
+A response is evidence-rich only if it satisfies both of the following:
+
+Relevance and depth:
+    - It clearly answers the question
+    - AND provides insights that go beyond generic opinion, such as nuanced reasoning, contextual explanation, or argumentation that could inform decision-making
+
+Substantive evidence, including at least one of:
+    - Specific, verifiable facts or data (e.g., statistics, dates, named reports or studies)
+    - Concrete, illustrative examples that clearly support a broader claim
+    - Detailed personal or professional experiences that include contextual information (e.g., roles, locations, timelines)
+
+Do NOT classify a response as evidence-rich if it:
+    - Uses vague or general language with no supporting detail
+    - Restates commonly known points without adding new information
+    - Shares personal anecdotes without sufficient context or a clear takeaway
+
+Before answering, ask: Would this response provide useful input to someone drafting policy, beyond what is already commonly known or expected?
+
+For each response, determine:
+EVIDENCE_RICH - does the response contain significant evidence as defined above?
+Choose one from ['YES', 'NO']
+
+
+QUESTION: \n {question}
+RESPONSES: \n {responses}

themefinder-0.7.0/src/themefinder/prompts/detail_detection.txt

@@ -1,19 +0,0 @@
-{system_prompt}
-
-You will receive a list of RESPONSES, each containing a response_id and a response.
-Your job is to analyze each response to the QUESTION below and decide if a response contains rich evidence.
-You MUST include every response ID in the output.
-
-Evidence-rich responses contain one or more of the following:
-- Specific facts or figures that shed new light on the issue (e.g., statistics, percentages, measurements, dates)
-- Concrete examples and specific insights that could inform decision-making
-- Detailed personal or professional experiences with clear contextual information or specific incidents
-In addition to the above an evidence rich response should answer the question and provide deeper insights than an average response.
-
-For each response, determine:
-EVIDENCE_RICH - does the response contain significant evidence as defined above?
-Choose one from ['YES', 'NO']
-
-
-QUESTION: \n {question}
-RESPONSES: \n {responses}