fix: align neuro-symbolic explanation and figure

docs/chapter1/Chapter1-Introduction-to-Agents.md

@@ -111,33 +111,34 @@ Through this approach, the agent decomposes a grand task requiring long-term pla
 
 This is a more fundamental classification dimension that explores what form the knowledge used by agents for decision-making exists in their "minds." This question is at the core of a debate that has lasted more than half a century in the field of artificial intelligence and has shaped two distinctly different AI cultures.
 
-- **Symbolic AI**
+- **Sub-symbolic AI**
 
-Symbolism, often called traditional artificial intelligence, has a core belief: intelligence stems from logical operations on symbols. The symbols here are human-readable entities (such as words, concepts), and operations follow strict logical rules, as shown on the left side of Figure 1.4. This is like a meticulous librarian organizing world knowledge into clear rule bases and knowledge graphs.
+Sub-symbolism, or connectionism, holds that knowledge does not take the form of explicit rules. Instead, it is implicitly distributed across a complex network of neurons as statistical patterns learned from massive amounts of data. Neural networks and deep learning are its representative approaches.
 
-Its main advantage lies in transparency and interpretability. Since reasoning steps are explicit, its decision-making process can be fully traced, which is crucial in high-risk fields such as finance and healthcare. However, its "Achilles' heel" lies in fragility: it relies on a complete rule system, but in the real world full of ambiguity and exceptions, any new situation not covered can lead to system failure, which is the so-called "knowledge acquisition bottleneck."
+As shown on the left side of Figure 1.4, sub-symbolic AI is like a babbling child. It does not learn to recognize cats from rules such as "cats have four legs, are furry, and meow." Instead, after seeing thousands of cat images, its neural network learns the visual patterns associated with the concept of a cat. This approach is powerful in pattern recognition and robust to noisy data. It can readily process unstructured data such as images and sounds, tasks that are often difficult for systems that rely on explicit rules.
 
-- **Sub-symbolic AI**
+However, this powerful intuitive capability also comes with opacity. Sub-symbolic systems are typically viewed as a **Black Box**. It can identify a cat in a picture with amazing accuracy, but if you ask it "why do you think this is a cat?", it likely cannot provide a logically sound explanation. Additionally, it performs poorly on pure logical reasoning tasks and sometimes produces hallucinations that seem reasonable but are factually incorrect.
 
-Sub-symbolism, or connectionism, provides a completely different picture. Here, knowledge is not explicit rules but implicitly distributed in a complex network composed of numerous neurons, representing statistical patterns learned from massive data. Neural networks and deep learning are its representatives.
+- **Symbolic AI**
 
-As shown in the middle of Figure 1.4, if symbolic AI is a librarian, then sub-symbolic AI is like a babbling child. They don't learn to recognize cats by learning rules like "cats have four legs, are furry, and meow," but after seeing thousands of cat pictures, the neural network in their brain can identify the visual pattern of the concept "cat." The power of this approach lies in its pattern recognition capability and robustness to noisy data. It can easily handle unstructured data such as images and sounds, which are extremely difficult tasks for symbolic AI.
+In contrast to sub-symbolic AI, symbolism, often called traditional artificial intelligence, holds that intelligence stems from logical operations on symbols. These symbols are human-readable entities such as words and concepts, and the operations follow strict logical rules, as shown in the middle of Figure 1.4. This is like a meticulous librarian organizing world knowledge into clear rule bases and knowledge graphs.
 
-However, this powerful intuitive capability also comes with opacity. Sub-symbolic systems are typically viewed as a **Black Box**. It can identify a cat in a picture with amazing accuracy, but if you ask it "why do you think this is a cat?", it likely cannot provide a logically sound explanation. Additionally, it performs poorly on pure logical reasoning tasks and sometimes produces hallucinations that seem reasonable but are factually incorrect.
+Its main advantage lies in transparency and interpretability. Since reasoning steps are explicit, its decision-making process can be fully traced, which is crucial in high-risk fields such as finance and healthcare. However, its "Achilles' heel" lies in fragility: it relies on a complete rule system, but in the real world full of ambiguity and exceptions, any new situation not covered can lead to system failure, which is the so-called "knowledge acquisition bottleneck."
 
 - **Neuro-Symbolic AI**
 
-For a long time, the two camps of symbolism and sub-symbolism developed like two parallel lines. To overcome the limitations of the above two paradigms, a "grand reconciliation" idea began to emerge, which is neuro-symbolic AI, also called neuro-symbolic hybrid. Its goal is to merge the advantages of both paradigms, creating a hybrid agent that can both learn from data like neural networks and perform logical reasoning like symbolic systems. It attempts to bridge the gap between perception and cognition, intuition and rationality. Nobel Prize-winning economist Daniel Kahneman's dual-system theory proposed in his book "Thinking, Fast and Slow" provides an excellent analogy for understanding neuro-symbolism<sup>[2]</sup>, as shown in Figure 1.4:
+For a long time, the two camps of sub-symbolism and symbolism developed like two parallel lines. To overcome the limitations of the above two paradigms, a "grand reconciliation" idea began to emerge, which is neuro-symbolic AI, also called neuro-symbolic hybrid. Its goal is to merge the advantages of both paradigms, creating a hybrid agent that can both learn from data like neural networks and perform logical reasoning like symbolic systems. In other words, it attempts to combine the pattern recognition strengths of sub-symbolic AI with the logical reasoning strengths of symbolic AI. Nobel Prize-winning economist Daniel Kahneman's dual-system theory proposed in his book "Thinking, Fast and Slow" provides an excellent analogy for understanding neuro-symbolism<sup>[2]</sup>. Following the left-to-right order in Figure 1.4, the correspondence can be summarized as follows:
 
 - **System 1** is a fast, intuitive, parallel thinking mode, similar to the powerful pattern recognition capability of sub-symbolic AI.
 - **System 2** is slow, methodical, logic-based deliberative thinking, just like the reasoning process of symbolic AI.
+- **Neuro-symbolic AI** combines the pattern recognition of System 1 with the logical reasoning of System 2, enabling the two to work together.
 
 <div align="center">
-  <img src="https://raw.githubusercontent.com/datawhalechina/Hello-Agents/main/docs/images/1-figures/1757242319667-4.png" alt="Figure description" width="90%"/>
-  <p>Figure 1.4 Knowledge representation paradigms of symbolism, sub-symbolism, and neuro-symbolic hybrid</p>
+  <img src="https://raw.githubusercontent.com/datawhalechina/Hello-Agents/main/docs/images/1-figures/1757242319667-4.png" alt="Relationship among sub-symbolic, symbolic, and neuro-symbolic AI" width="90%"/>
+  <p>Figure 1.4 Knowledge representation paradigms of sub-symbolism, symbolism, and neuro-symbolic hybrid</p>
 </div>
 
-Human intelligence stems from the collaborative work of these two systems. Similarly, a truly robust AI also needs to combine the strengths of both. Large language model-driven agents are an excellent practical example of neuro-symbolism. Its core is a huge neural network, giving it pattern recognition and language generation capabilities. However, when it works, it generates a series of structured intermediate steps, such as thoughts, plans, or API calls, which are all explicit, operable symbols. Through this approach, it achieves a preliminary fusion of perception and cognition, intuition and rationality.
+Human intelligence stems from the collaborative work of these two systems. Similarly, a truly robust AI also needs to combine the strengths of both. Large language model-driven agents are an excellent practical example of neuro-symbolism: their core is a large neural network that provides pattern recognition and language generation capabilities; during operation, they also generate a series of structured intermediate steps, such as thoughts, plans, or API calls, which are explicit, operable symbols. In this way, they combine neural pattern recognition with symbolic logical reasoning.
 
 ## 1.2 Composition and Operating Principles of Agents
 

docs/chapter1/第一章 初识智能体.md

@@ -112,33 +112,34 @@
 
 这是一个更根本的分类维度，它探究智能体用以决策的知识，究竟是以何种形式存于其“思想”之中。这个问题是人工智能领域一场持续半个多世纪的辩论核心，并塑造了两种截然不同的 AI 文化。
 
-- <strong>符号主义 AI（Symbolic AI）</strong>
+- <strong>亚符号主义 AI（Sub-symbolic AI）</strong>
 
-符号主义，常被称为传统人工智能，其核心信念是：智能源于对符号的逻辑操作。这里的符号是人类可读的实体（如词语、概念），操作则遵循严格的逻辑规则，如图 1.4 左侧所示。这好比一位一丝不苟的图书管理员，将世界知识整理为清晰的规则库和知识图谱。
+亚符号主义，或称连接主义，认为知识并非显式的规则，而是内隐地分布在一个由大量神经元组成的复杂网络中，是从海量数据中学习到的统计模式。神经网络和深度学习是其代表。
 
-其主要优势在于透明和可解释。由于推理步骤明确，其决策过程可以被完整追溯，这在金融、医疗等高风险领域至关重要。然而，其“阿喀琉斯之踵”在于脆弱性：它依赖于一个完备的规则体系，但在充满模糊和例外的现实世界中，任何未被覆盖的新情况都可能导致系统失灵，这就是所谓的“知识获取瓶颈”。
+如图 1.4 左侧所示，亚符号主义 AI 就像一个牙牙学语的孩童。他不是通过学习“猫有四条腿、毛茸茸、会喵喵叫”这样的规则来认识猫的，而是在看过成千上万张猫的图片后，大脑中的神经网络能辨识出“猫”这个概念的视觉模式。这种方法的强大之处在于其模式识别能力和对噪声数据的鲁棒性。它能够轻松处理图像、声音等非结构化数据，而这类任务对于依赖明确规则的系统往往非常困难。
 
-- <strong>亚符号主义 AI（Sub-symbolic AI）</strong>
+然而，这种强大的直觉能力也伴随着不透明性。亚符号主义系统通常被视为一个<strong>黑箱（Black Box）</strong>。它能以惊人的准确率识别出图片中的猫，但你若问它“为什么你认为这是猫？”，它很可能无法给出一个合乎逻辑的解释。此外，它在纯粹的逻辑推理任务上表现不佳，有时会产生看似合理却事实错误的幻觉。
 
-亚符号主义，或称连接主义，则提供了一幅截然不同的图景。在这里，知识并非显式的规则，而是内隐地分布在一个由大量神经元组成的复杂网络中，是从海量数据中学习到的统计模式。神经网络和深度学习是其代表。
+- <strong>符号主义 AI（Symbolic AI）</strong>
 
-如图 1.4 中间所示，如果说符号主义 AI 是图书管理员，那么亚符号主义 AI 就像一个牙牙学语的孩童 。他不是通过学习“猫有四条腿、毛茸茸、会喵喵叫”这样的规则来认识猫的，而是在看过成千上万张猫的图片后，大脑中的神经网络能辨识出“猫”这个概念的视觉模式 。这种方法的强大之处在于其模式识别能力和对噪声数据的鲁棒性 。它能够轻松处理图像、声音等非结构化数据，这在符号主义 AI 看来是极其困难的任务。
+与亚符号主义不同，符号主义常被称为传统人工智能，其核心信念是：智能源于对符号的逻辑操作。这里的符号是人类可读的实体（如词语、概念），操作则遵循严格的逻辑规则，如图 1.4 中间所示。这好比一位一丝不苟的图书管理员，将世界知识整理为清晰的规则库和知识图谱。
 
-然而，这种强大的直觉能力也伴随着不透明性。亚符号主义系统通常被视为一个<strong>黑箱（Black Box）</strong>。它能以惊人的准确率识别出图片中的猫，但你若问它“为什么你认为这是猫？”，它很可能无法给出一个合乎逻辑的解释。此外，它在纯粹的逻辑推理任务上表现不佳，有时会产生看似合理却事实错误的幻觉 。
+其主要优势在于透明和可解释。由于推理步骤明确，其决策过程可以被完整追溯，这在金融、医疗等高风险领域至关重要。然而，其“阿喀琉斯之踵”在于脆弱性：它依赖于一个完备的规则体系，但在充满模糊和例外的现实世界中，任何未被覆盖的新情况都可能导致系统失灵，这就是所谓的“知识获取瓶颈”。
 
 - <strong>神经符号主义 AI（Neuro-Symbolic AI）</strong>
 
-长久以来，符号主义和亚符号主义这两大阵营如同两条平行线，各自发展。为克服上述两种范式的局限，一种“大和解”的思想开始兴起，这就是神经符号主义 AI，也称神经符号混合主义。它的目标，是融合两大范式的优点，创造出一个既能像神经网络一样从数据中学习，又能像符号系统一样进行逻辑推理的混合智能体。它试图弥合感知与认知、直觉与理性之间的鸿沟。诺贝尔经济学奖得主丹尼尔·卡尼曼（Daniel Kahneman）在其著作《思考，快与慢》（Thinking, Fast and Slow）中提出的双系统理论，为我们理解神经符号主义提供了一个绝佳的类比<sup>[2]</sup>，如图 1.4 所示：
+长久以来，亚符号主义和符号主义这两大阵营如同两条平行线，各自发展。为克服上述两种范式的局限，一种“大和解”的思想开始兴起，这就是神经符号主义 AI，也称神经符号混合主义。它的目标，是融合两大范式的优点，创造出一个既能像神经网络一样从数据中学习，又能像符号系统一样进行逻辑推理的混合智能体。换言之，它试图将亚符号主义擅长的模式识别与符号主义擅长的逻辑推理结合起来。诺贝尔经济学奖得主丹尼尔·卡尼曼（Daniel Kahneman）在其著作《思考，快与慢》（Thinking, Fast and Slow）中提出的双系统理论，为我们理解神经符号主义提供了一个绝佳的类比<sup>[2]</sup>。按照图 1.4 从左到右的展示顺序，这种对应关系可以概括为：
 
 - <strong>系统 1</strong>是快速、凭直觉、并行的思维模式，类似于亚符号主义 AI 强大的模式识别能力。
 - <strong>系统 2</strong>是缓慢、有条理、基于逻辑的审慎思维，恰如符号主义 AI 的推理过程。
+- <strong>神经符号主义 AI</strong>则将系统 1 的模式识别与系统 2 的逻辑推理结合起来，使二者协同工作。
 
 <div align="center">
-  <img src="https://raw.githubusercontent.com/datawhalechina/Hello-Agents/main/docs/images/1-figures/1757242319667-4.png" alt="图片描述" width="90%"/>
-  <p>图 1.4 符号主义、亚符号主义与神经符号混合主义的知识表示范式</p>
+  <img src="https://raw.githubusercontent.com/datawhalechina/Hello-Agents/main/docs/images/1-figures/1757242319667-4.png" alt="亚符号主义、符号主义与神经符号主义的关系示意图" width="90%"/>
+  <p>图 1.4 亚符号主义、符号主义与神经符号混合主义的知识表示范式</p>
 </div>
 
-人类的智能，正源于这两个系统的协同工作。同样，一个真正鲁棒的 AI，也需要兼具二者之长。大语言模型驱动的智能体是神经符号主义的一个极佳实践范例。其内核是一个巨大的神经网络，使其具备模式识别和语言生成能力。然而，当它工作时，它会生成一系列结构化的中间步骤，如思想、计划或 API 调用，这些都是明确的、可操作的符号。通过这种方式，它实现了感知与认知、直觉与理性的初步融合。
+人类的智能，正源于这两个系统的协同工作。同样，一个真正鲁棒的 AI，也需要兼具二者之长。大语言模型驱动的智能体是神经符号主义的一个极佳实践范例：其内核是一个巨大的神经网络，使其具备模式识别和语言生成能力；在工作过程中，它又会生成一系列结构化的中间步骤，如思想、计划或 API 调用，这些都是明确的、可操作的符号。通过这种方式，它将基于神经网络的模式识别与基于符号的逻辑推理结合起来。