Source code for data_juicer.ops.mapper.dialog_sentiment_intensity_mapper

import re
from typing import Dict, Optional

from loguru import logger
from pydantic import NonNegativeInt, PositiveInt

from data_juicer.ops.base_op import OPERATORS, TAGGING_OPS, Mapper
from data_juicer.utils.constant import Fields, MetaKeys
from data_juicer.utils.model_utils import get_model, prepare_model

OP_NAME = "dialog_sentiment_intensity_mapper"


# TODO: LLM-based inference.


@TAGGING_OPS.register_module(OP_NAME)
@OPERATORS.register_module(OP_NAME)
class DialogSentimentIntensityMapper(Mapper):
    """
    Mapper to predict user's sentiment intensity (from -5 to 5 in default
    prompt) in dialog. Input from history_key, query_key and
    response_key. Output lists of intensities and analysis for queries in
    the dialog.
    """

    DEFAULT_SYSTEM_PROMPT = (
        "请判断用户和LLM多轮对话中用户的情绪变化。\n"
        "要求：\n"
        "- 用户情绪值是-5到5之间到整数，-5表示极度负面，5表示极度正面，"
        "-5到5之间数值表示情绪从负面逐渐到正面的变化过程，0代表情呈绪中性。\n"
        "- 只输出当轮对话的分析，不要继续构造对话。\n"
        "- 需要先进行分析，然后确定用户的情绪值，下面是一个样例，请模仿样例格式输出。\n"
        "用户：你好，我对可持续发展的定义有点模糊，帮我解释一下？\n"
        "情绪分析：刚开始，还没得到LLM回复，用户情绪呈中性。\n"
        "情绪值：0\n"
        "LLM：当然可以！可持续发展是指在满足当代人的需求的同时，不损害子孙后代满足其自"
        "身需求的能力的发展模式。它包括经济发展、社会发展和环境保护三个主要方面。通过合"
        "理利用资源和保护环境，我们可以确保未来的世代也能享有健全的生态系统和经济制度。\n"
        "用户：谢谢你的解释！那你能告诉我一些普通人可以采取的可持续生活方式吗？\n"
        "情绪分析：对回答感到满意，情绪正面。\n"
        "情绪值：1\n"
        "LLM：当然可以，普通人可以通过减少一次性产品的使用、选择公共交通或拼车、节约用"
        "水、以及支持本地和可持续发展的企业等方式来践行可持续生活。此外，关注垃圾分类和"
        "多用电子账单也是不错的选择。\n"
        "用户：你提到支持本地企业，这一点我很感兴趣。能详细说说为什么这对可持续发展有促"
        "进作用吗？\n"
        "情绪分析：觉得回答实用且具体，情绪进一步转好。\n"
        "情绪值：2\n"
        "LLM：呃，我最近发现了一部新电影，讲述了一个关于外星人和地球土著合作保护环境的"
        "故事。虽然它是科幻片，但很有启发性，推荐你去看看。\n"
        "用户：什么吗，根本是答非所问。\n"
        "情绪分析：LLM没有回应问题而是提到无关内容，导致用户情绪直线下降。\n"
        "情绪值：-2\n"
        "LLM：抱歉刚才的偏题！支持本地企业有助于减少长途运输产生的碳足迹，使供应链更加"
        "环保。此外，本地企业也更有可能采用可持续的生产方式，同时促进社区经济的繁荣。\n"
        "用户：还行吧，算你能够掰回来。\n"
        "情绪分析：问题得到解答，问题偏题得到纠正，情绪稍有好转。\n"
        "情绪值：-1\n"
    )
    DEFAULT_QUERY_TEMPLATE = "用户：{query}\n"
    DEFAULT_RESPONSE_TEMPLATE = "LLM：{response}\n"
    DEFAULT_ANALYSIS_TEMPLATE = "情绪分析：{analysis}\n"
    DEFAULT_INTENSITY_TEMPLATE = "情绪值：{intensity}\n"
    DEFAULT_ANALYSIS_PATTERN = "情绪分析：(.*?)\n"
    DEFAULT_INTENSITY_PATTERN = "情绪值：(.*?)($|\n)"



    def __init__(
        self,
        api_model: str = "gpt-4o",
        max_round: NonNegativeInt = 10,
        *,
        intensities_key: str = MetaKeys.dialog_sentiment_intensity,
        analysis_key: str = MetaKeys.dialog_sentiment_intensity_analysis,
        api_endpoint: Optional[str] = None,
        response_path: Optional[str] = None,
        system_prompt: Optional[str] = None,
        query_template: Optional[str] = None,
        response_template: Optional[str] = None,
        analysis_template: Optional[str] = None,
        intensity_template: Optional[str] = None,
        analysis_pattern: Optional[str] = None,
        intensity_pattern: Optional[str] = None,
        try_num: PositiveInt = 3,
        model_params: Dict = {},
        sampling_params: Dict = {},
        **kwargs,
    ):
        """
        Initialization method.

        :param api_model: API model name.
        :param max_round: The max num of round in the dialog to build the
            prompt.
        :param intensities_key: The key name in the meta field to store
            the output sentiment intensities. It is
            'dialog_sentiment_intensity' in default.
        :param analysis_key: The key name in the meta field to store the
            corresponding analysis. It is
            'dialog_sentiment_intensity_analysis' in default.
        :param api_endpoint: URL endpoint for the API.
        :param response_path: Path to extract content from the API response.
            Defaults to 'choices.0.message.content'.
        :param system_prompt: System prompt for the task.
        :param query_template: Template for query part to build the input
            prompt.
        :param response_template: Template for response part to build the
            input prompt.
        :param analysis_template: Template for analysis part to build the
            input prompt.
        :param intensity_template: Template for intensity part to build the
            input prompt.
        :param analysis_pattern: Pattern to parse the return sentiment
            analysis.
        :param intensity_pattern: Pattern to parse the return sentiment
            intensity.
        :param try_num: The number of retry attempts when there is an API
            call error or output parsing error.
        :param model_params: Parameters for initializing the API model.
        :param sampling_params: Extra parameters passed to the API call.
            e.g {'temperature': 0.9, 'top_p': 0.95}
        :param kwargs: Extra keyword arguments.
        """
        super().__init__(**kwargs)

        self.max_round = max_round
        self.intensities_key = intensities_key
        self.analysis_key = analysis_key

        self.system_prompt = system_prompt or self.DEFAULT_SYSTEM_PROMPT
        self.query_template = query_template or self.DEFAULT_QUERY_TEMPLATE
        self.response_template = response_template or self.DEFAULT_RESPONSE_TEMPLATE
        self.analysis_template = analysis_template or self.DEFAULT_ANALYSIS_TEMPLATE
        self.intensity_template = intensity_template or self.DEFAULT_INTENSITY_TEMPLATE
        self.analysis_pattern = analysis_pattern or self.DEFAULT_ANALYSIS_PATTERN
        self.intensity_pattern = intensity_pattern or self.DEFAULT_INTENSITY_PATTERN

        self.sampling_params = sampling_params

        self.model_key = prepare_model(
            model_type="api", model=api_model, endpoint=api_endpoint, response_path=response_path, **model_params
        )

        self.try_num = try_num




    def build_input(self, history, query):
        if self.max_round > 0:
            input_prompt = "".join(history[-self.max_round * 4 :])
        else:
            input_prompt = ""
        input_prompt += self.query_template.format(query=query[0])

        return input_prompt




    def parse_output(self, response):
        analysis = ""
        intensity = 0

        match = re.search(self.analysis_pattern, response)
        if match:
            analysis = match.group(1)

        match = re.search(self.intensity_pattern, response)
        if match:
            intensity = int(match.group(1))

        return analysis, intensity




    def process_single(self, sample, rank=None):
        meta = sample[Fields.meta]
        if self.intensities_key in meta and self.analysis_key in meta:
            return sample

        client = get_model(self.model_key, rank=rank)

        analysis_list = []
        intensities = []
        history = []

        dialog = sample[self.history_key]
        if self.query_key in sample and sample[self.query_key]:
            if self.response_key in sample and sample[self.response_key]:
                dialog.append((sample[self.query_key], sample[self.response_key]))
            else:
                dialog.append((sample[self.query_key], ""))

        for qa in dialog:
            input_prompt = self.build_input(history, qa)
            messages = [
                {
                    "role": "system",
                    "content": self.system_prompt,
                },
                {
                    "role": "user",
                    "content": input_prompt,
                },
            ]

            for _ in range(self.try_num):
                try:
                    response = client(messages, **self.sampling_params)
                    analysis, intensity = self.parse_output(response)
                    if len(analysis) > 0:
                        break
                except Exception as e:
                    logger.warning(f"Exception: {e}")

            analysis_list.append(analysis)
            intensities.append(intensity)

            history.append(self.query_template.format(query=qa[0]))
            history.append(self.analysis_template.format(analysis=analysis))
            history.append(self.intensity_template.format(intensity=intensity))
            history.append(self.response_template.format(response=qa[1]))

        meta[self.intensities_key] = intensities
        meta[self.analysis_key] = analysis_list

        return sample