正向链接：
前向链接从可用数据和用户推理规则开始，从最终用户那里提取更多数据，直到达到目标。推理应用于此信息以获得合乎逻辑的结论。它是一种给定一个或多个条件的系统，其中系统搜索和规则知识库选择每个条件和与条件相对应的规则，规则可以从结论中产生新的条件。其中事实保存在工作记忆中，条件动作规则表示当特定事实出现在工作记忆中时要采取的动作，它可以从工作记忆中添加和删除事实。
反向链接：
它基于获取初始状态的决定，然后称为反向链接。反向链接也称为决策驱动或目标驱动的推理技术。在反向链接中，推理是在反向进行的，系统在反向选择目标状态和推理。其中我们选择一个目标状态和规则，目标状态作为结论建立子目标满足所有目标状态为真，并且有一个条件满足所有子目标并检查给定的初始状态是否匹配与既定的。
| Forward Chaining | Backward chaining |
|---|---|
| Forward chaining suitable for breadth first search. | Backward chaining suitable for depth search. |
| It begins with initial facts. | It begins with some hypothesis goal. |
| It may slow, because in which we tested all the rules. | It may fast as compared to Forward chaining because it test fewer rules. |
| It provides small amount of data in which we use to store large amount of information. | It provides small amount of data in which we store small information. |
| It is basically on primarily data driven. | It is basically on goal driven. |
| It follows Top down reasoning. | It follows bottom-up reasoning. |
| It contains small number of initial states but large number of conclusion. | It contains small number of initial goals and large number of rules. |
| It is suitable for data collection problem like planning monitoring. | It is suitable for hypothesis problem like diagnosis. |
| In which all data is available. | In which data must be acquired. |