Release new docs

Author: Milvus-doc-bot

v0.10.2/assets/IP.png

@@ -27,20 +27,27 @@ It's the most commonly used distance metric, and is very useful when the data is
 
 ### Inner product (IP)
 
-IP measures the cosine of the angle between 2 vectors, and returns the normalized dot product of them.
-
-So the formula for IP is:
+The IP distance between two embeddings are defined as follows: 
 
 ![ip](../../../assets/ip_metric.png)
 
-where A and B are vectors, `||A||` and `||B||` are the norms of A and B, and cosθ is the cosine of the angle between A and B.
+where A and B are embeddings, `||A||` and `||B||` are the norms of A and B.
 
 IP is more useful if you are more interested in measuring the orientation but not the magnitude of the vectors.
 
 <div class="alert note">
-If the vectors are normalized, IP is equivalent to cosine similarity. Thus, Milvus does not provide a metric for cosine similarity.
+ If you use IP to calculate embeddings similarities, you must normalize your embeddings. After normalization, inner product equals cosine similarity.
 </div>
 
+
+Suppose X' is normalized from embedding X: 
+
+![normalize](../../../assets/normalize.png)
+
+The correlation between the two embeddings is as follows: 
+
+![normalization](../../../assets/normalization.png)
+
 ### Jaccard distance
 
 Jaccard similarity coefficient measures the similarity between two sample sets, and is defined as the cardinality of the intersection of the defined sets divided by the cardinality of the union of them. It can only be applied to finite sample sets.

v0.10.2/assets/normalization.png

@@ -27,20 +27,27 @@ Milvus 基于不同的距离计算方式比较向量间的距离。选择合适
 
 ### 内积 （IP）
 
-内积计算两条向量之间的夹角余弦，并返回相应的点积。
+两条向量内积距离的计算公式为：
 
-内积距离的计算公式为：
+![ip](../../../assets/IP.png)
 
-![ip](../../../assets/ip_metric.png)
 
-假设有 A 和 B 两条向量，则 `||A||` 与 `||B||` 分别代表 A 和 B 归一化后的值。cosθ 代表 A 与 B 之间的余弦夹角。
+假设有 A 和 B 两条向量，则 `||A||` 与 `||B||` 分别代表 A 和 B 归一化后的值。
 
 内积更适合计算向量的方向而不是大小。
 
 <div class="alert note">
-在向量归一化之后，内积与余弦相似度等价。因此 Milvus 并没有单独提供余弦相似度作为向量距离计算方式。
+如需使用点积计算向量相似度，则必须对向量作归一化处理。处理后点积与余弦相似度等价。
 </div>
 
+假设 X' 是向量 X 的归一化向量：
+
+![normalize](../../../assets/normalize.png)
+
+两者之间的关系为：
+
+![normalization](../../../assets/normalization.png)
+
 ### 杰卡德距离
 
 杰卡德相似系数计算数据集之间的相似度，计算方式为：数据集交集的个数和并集个数的比值。计算公式可以表示为：