c7d259d5fd
This PR ensures that only conversations from quick conversation channels are resolved, avoiding resolutions on the email channel (we still need to improve the UX here). It also updates the FAQ generation logic, limiting it to conversations that had at least one human interaction.
119 lines
3.1 KiB
Ruby
119 lines
3.1 KiB
Ruby
class Captain::Llm::ConversationFaqService < Captain::Llm::BaseOpenAiService
|
|
DISTANCE_THRESHOLD = 0.3
|
|
|
|
def initialize(assistant, conversation)
|
|
super()
|
|
@assistant = assistant
|
|
@conversation = conversation
|
|
@content = conversation.to_llm_text
|
|
end
|
|
|
|
# Generates and deduplicates FAQs from conversation content
|
|
# Skips processing if there was no human interaction
|
|
def generate_and_deduplicate
|
|
return [] if no_human_interaction?
|
|
|
|
new_faqs = generate
|
|
return [] if new_faqs.empty?
|
|
|
|
duplicate_faqs, unique_faqs = find_and_separate_duplicates(new_faqs)
|
|
save_new_faqs(unique_faqs)
|
|
log_duplicate_faqs(duplicate_faqs) if Rails.env.development?
|
|
end
|
|
|
|
private
|
|
|
|
attr_reader :content, :conversation, :assistant
|
|
|
|
def no_human_interaction?
|
|
conversation.first_reply_created_at.nil?
|
|
end
|
|
|
|
def find_and_separate_duplicates(faqs)
|
|
duplicate_faqs = []
|
|
unique_faqs = []
|
|
|
|
faqs.each do |faq|
|
|
combined_text = "#{faq['question']}: #{faq['answer']}"
|
|
embedding = Captain::Llm::EmbeddingService.new.get_embedding(combined_text)
|
|
similar_faqs = find_similar_faqs(embedding)
|
|
|
|
if similar_faqs.any?
|
|
duplicate_faqs << { faq: faq, similar_faqs: similar_faqs }
|
|
else
|
|
unique_faqs << faq
|
|
end
|
|
end
|
|
|
|
[duplicate_faqs, unique_faqs]
|
|
end
|
|
|
|
def find_similar_faqs(embedding)
|
|
similar_faqs = assistant
|
|
.responses
|
|
.nearest_neighbors(:embedding, embedding, distance: 'cosine')
|
|
Rails.logger.debug(similar_faqs.map { |faq| [faq.question, faq.neighbor_distance] })
|
|
similar_faqs.select { |record| record.neighbor_distance < DISTANCE_THRESHOLD }
|
|
end
|
|
|
|
def save_new_faqs(faqs)
|
|
faqs.map do |faq|
|
|
assistant.responses.create!(
|
|
question: faq['question'],
|
|
answer: faq['answer'],
|
|
status: 'pending',
|
|
documentable: conversation
|
|
)
|
|
end
|
|
end
|
|
|
|
def log_duplicate_faqs(duplicate_faqs)
|
|
return if duplicate_faqs.empty?
|
|
|
|
Rails.logger.info "Found #{duplicate_faqs.length} duplicate FAQs:"
|
|
duplicate_faqs.each do |duplicate|
|
|
Rails.logger.info(
|
|
"Q: #{duplicate[:faq]['question']}\n" \
|
|
"A: #{duplicate[:faq]['answer']}\n\n" \
|
|
"Similar existing FAQs: #{duplicate[:similar_faqs].map { |f| "Q: #{f.question} A: #{f.answer}" }.join(', ')}"
|
|
)
|
|
end
|
|
end
|
|
|
|
def generate
|
|
response = @client.chat(parameters: chat_parameters)
|
|
parse_response(response)
|
|
rescue OpenAI::Error => e
|
|
Rails.logger.error "OpenAI API Error: #{e.message}"
|
|
[]
|
|
end
|
|
|
|
def chat_parameters
|
|
prompt = Captain::Llm::SystemPromptsService.conversation_faq_generator
|
|
{
|
|
model: @model,
|
|
response_format: { type: 'json_object' },
|
|
messages: [
|
|
{
|
|
role: 'system',
|
|
content: prompt
|
|
},
|
|
{
|
|
role: 'user',
|
|
content: content
|
|
}
|
|
]
|
|
}
|
|
end
|
|
|
|
def parse_response(response)
|
|
content = response.dig('choices', 0, 'message', 'content')
|
|
return [] if content.nil?
|
|
|
|
JSON.parse(content.strip).fetch('faqs', [])
|
|
rescue JSON::ParserError => e
|
|
Rails.logger.error "Error in parsing GPT processed response: #{e.message}"
|
|
[]
|
|
end
|
|
end
|