Purpose

Current possibilities of accessing business data by regular users usually involve complicated user interfaces or require technical expertise. This results in situations when business owners are separated from their data. The aim of this research is to apply an innovative approach leveraging conversational interfaces to tackle this problem.

Design/methodology/approach

The authors examine the current possibilities of accessing business data by business, users with an emphasis on conversational interfaces employing a chatbot as an alternative to traditional approaches. The authors propose a new concept relying on a guided conversation, and through experiments with a real chatbot and database, the authors demonstrate the benefits of the proposed approach.

Findings

The authors found out that the key to the success of our approach is a decomposition of complex database queries and their incremental construction in conversations. This also enables natural discovery of the domain model through constantly provided feedback. Based on the experiments with a real chatbot, the authors demonstrate that defining conversation flows and maintaining the conversation context is a crucial aspect contributing to the overall accuracy, together with keeping the conversation within the defined limits in its certain parts.

Originality/value

The authors present a novel approach using natural language interfaces for accessing data by business users. In contrast to existing approaches, the authors emphasize incremental construction of queries, predefined conversation flows and constraining the conversations, when necessary.

中文翻译：

---

用于非常规访问业务关系数据结构的会话界面

目的

当前由普通用户访问业务数据的可能性通常涉及复杂的用户界面或需要技术专长。这会导致业务所有者与其数据分离的情况。这项研究的目的是应用一种利用对话界面的创新方法来解决这个问题。

设计/方法/方法

作者研究了当前按业务访问业务数据的可能性，用户强调会话界面，使用聊天机器人作为传统方法的替代方案。作者提出了一个依赖于引导对话的新概念，并通过对真实聊天机器人和数据库的实验，证明了所提出方法的好处。

发现

作者发现，我们方法成功的关键是复杂数据库查询的分解及其在对话中的增量构造。这还可以通过不断提供的反馈来自然地发现领域模型。基于对真实聊天机器人的实验，作者证明了定义对话流和维护对话上下文是有助于整体准确性的关键方面，同时将对话保持在其某些部分的定义范围内。

原创性/价值

作者提出了一种使用自然语言界面来访问业务用户数据的新颖方法。与现有方法相比，作者强调查询的增量构造、预定义的对话流以及在必要时限制对话。