diff --git a/blog/en/2020-07-28-Building-an-AI-Powered-Writing-Assistant-with-WPS-Office.md b/blog/en/2020-07-28-Building-an-AI-Powered-Writing-Assistant-with-WPS-Office.md
index 07d304b1..01a11cf0 100644
--- a/blog/en/2020-07-28-Building-an-AI-Powered-Writing-Assistant-with-WPS-Office.md
+++ b/blog/en/2020-07-28-Building-an-AI-Powered-Writing-Assistant-with-WPS-Office.md
@@ -59,7 +59,7 @@ The neural network is now capable of extracting and classifying named entities f
 
 [Milvus](https://www.milvus.io/) is an open source similarity search engine that supports adding, deleting, updating, and near-real-time search of embeddings on a trillion bytes scale. To improve query performance, Milvus allows an index type to be specified for each vector field. The WPS Office smart assistant uses the IVF_FLAT index, the most basic Inverted File (IVF) index type where “flat” means vectors are stored without compression or quantization. Clustering is based on IndexFlat2, which uses exact search for L2 distance.
 
-Although IVF_FLAT has a 100% query recall rate, its lack of compression results in comparatively slow query speeds. Milvus’ [partitioning function](https://www.milvus.io/docs/v0.10.2/storage_concept.md) is used to divide data into multiple parts of physical storage based on predefined rules, making queries faster and more accurate. When vectors are added to Milvus, tags specify which partition the data should be added to. Queries of the vector data use tags to specify which partition the query should be executed on. Data can be further broken down into segments within each partition to further improve speed.
+Although IVF_FLAT has a 100% query recall rate, its lack of compression results in comparatively slow query speeds. Milvus’ [partitioning function](https://milvus.io/docs/manage-partitions.md) is used to divide data into multiple parts of physical storage based on predefined rules, making queries faster and more accurate. When vectors are added to Milvus, tags specify which partition the data should be added to. Queries of the vector data use tags to specify which partition the query should be executed on. Data can be further broken down into segments within each partition to further improve speed.
 
 The intelligent writing assistant also uses Kubernetes clusters, allowing application containers to run across multiple machines and environments, as well as MySQL for metadata management.