3590 - 使用基于LLM的语义增强将业务上下文应用于表和列.pdf

《3590 - 使用基于LLM的语义增强将业务上下文应用于表和列.pdf》由会员分享，可在线阅读，更多相关《3590 - 使用基于LLM的语义增强将业务上下文应用于表和列.pdf（18页珍藏版）》请在三个皮匠报告上搜索。

1、Semantic Enrichments using LLMs-Sharath Aithal(Sofware Development Manager)Why do you need a Semantic Layer?It provides a simplified view of complex data structures,making it easier for users to discover,understand and interact with the data.Data Abstraction The semantic layer ensures that data is c

2、onsistently defined and interpreted across the organization,reducing confusion and miscommunication.Consistent Data DefinitionsUsers can create queries using familiar business terminology rather than technical database language,streamlining the data analysis process.Enhanced Querying By defining cle

3、ar relationships and rules within the data,the semantic layer aids in enforcing data governance policies and ensures compliance.Improved Data GovernanceIt allows different tools and systems to communicate more effectively,facilitating integration and data sharing across platforms.Interoperability Bu

4、siness users can derive insights more independently,reducing the reliance on data engineers or IT teams for accessing and interpreting data.User EmpowermentColumn Name generationSemantic name expansions depends on the pre-defined glossary.It uses fuzzy matching library to expand cryptic column names

5、 based on the given glossary into easy to understand namesColumn Description generationSemantic description generation uses a granite-8b-code-instruct-v2 model to generate descriptions.The model considers the context of the columns to provide more context aware descriptions.Glossary Generation and C

6、olumn Name Generation Using LLMs:The LLMs are used to generate glossary concepts from the metadata of the tables and columns.The LLMs are also used to expand the abbreviated column names.The additional information like sample values captured during profiling are leveraged to boost the accuracy of th