Enhancement for knowledge pre-processing components/知识加工组件贡献 #258
Description
======================== English version ==============================
The agentUniverse currently provides default knowledge pre-processing components, and the project requires other processing methods. We look forward to the following directions for PR:
Knowledge Fragmentation: Offer the most suitable knowledge fragmentation component based on specific task scenarios. For instance, in knowledge processing related to contract clauses, it is often necessary to fragment based on individual clauses; in academic paper processing, fragmentation might be based on arguments and evidence.
Knowledge Extraction: Provide the most appropriate knowledge extraction component based on specific task scenarios. For example, in supply chain knowledge scenarios, it is often crucial to extract entities and their relationships; in financial report scenarios, the extraction of corresponding financial indicators is typically needed.
Knowledge Deduplication: Offer the most suitable knowledge deduplication component based on specific task scenarios, including knowledge aggregation, summarization, and synthesis. For instance, in financial event scenarios, it is necessary to aggregate associated news articles related to the same event into a single event report.
======================== 中文版 ==============================
agentUniverse目前提供默认的知识加工组件，项目需要其他的加工手段，期待您建设的方向如下：
- 知识拆条：需要结合具体任务场景提供最合适的知识拆条组件。例如在合同条款类知识加工中，往往按照单条条款进行拆条；在论文类知识加工中，往往按照论点、论据进行拆条；
- 知识提取：需要结合具体任务场景提供最合适的知识拆条组件。例如，在供应链场景知识中往往会提取知识中的实体与其关联关系；在金融财报场景知识中往往会提取对应的金融指标；
- 知识去重：需要结合具体任务场景提供最合适的知识去重组件，包括对于知识的聚合、摘要、总结。例如，在金融事件场景中，需要将同一个事件的关联新闻聚合到同一个事件中去。