diff --git a/internal/core/thirdparty/tantivy/tantivy-binding/src/analyzer/filter/filter.rs b/internal/core/thirdparty/tantivy/tantivy-binding/src/analyzer/filter/filter.rs
index 05cbb30705..75f2328817 100644
--- a/internal/core/thirdparty/tantivy/tantivy-binding/src/analyzer/filter/filter.rs
+++ b/internal/core/thirdparty/tantivy/tantivy-binding/src/analyzer/filter/filter.rs
@@ -2,7 +2,7 @@ use serde_json as json;
 use tantivy::tokenizer::*;
 
 use super::util::*;
-use super::{RegexFilter, RemovePunctFilter};
+use super::{RegexFilter, RemovePunctFilter, SynonymFilter};
 use crate::error::{Result, TantivyBindingError};
 
 pub(crate) enum SystemFilter {
@@ -18,6 +18,7 @@ pub(crate) enum SystemFilter {
     Decompounder(SplitCompoundWords),
     Stemmer(Stemmer),
     Regex(RegexFilter),
+    Synonym(SynonymFilter),
 }
 
 impl SystemFilter {
@@ -34,6 +35,7 @@ impl SystemFilter {
             Self::Stemmer(filter) => builder.filter(filter).dynamic(),
             Self::RemovePunct(filter) => builder.filter(filter).dynamic(),
             Self::Regex(filter) => builder.filter(filter).dynamic(),
+            Self::Synonym(filter) => builder.filter(filter).dynamic(),
             Self::Invalid => builder,
         }
     }
@@ -182,6 +184,7 @@ impl TryFrom<&json::Map<String, json::Value>> for SystemFilter {
                     "decompounder" => get_decompounder_filter(params),
                     "stemmer" => get_stemmer_filter(params),
                     "regex" => RegexFilter::from_json(params).map(|f| SystemFilter::Regex(f)),
+                    "synonym" => SynonymFilter::from_json(params).map(|f| SystemFilter::Synonym(f)),
                     other => Err(TantivyBindingError::InternalError(format!(
                         "unsupport filter type: {}",
                         other
diff --git a/internal/core/thirdparty/tantivy/tantivy-binding/src/analyzer/filter/mod.rs b/internal/core/thirdparty/tantivy/tantivy-binding/src/analyzer/filter/mod.rs
index 54ad1636e8..b911566e5f 100644
--- a/internal/core/thirdparty/tantivy/tantivy-binding/src/analyzer/filter/mod.rs
+++ b/internal/core/thirdparty/tantivy/tantivy-binding/src/analyzer/filter/mod.rs
@@ -2,10 +2,12 @@ mod filter;
 mod regex_filter;
 mod remove_punct_filter;
 pub(crate) mod stop_words;
+mod synonym_filter;
 mod util;
 
 use regex_filter::RegexFilter;
 use remove_punct_filter::RemovePunctFilter;
+use synonym_filter::SynonymFilter;
 
 pub(crate) use filter::*;
 pub(crate) use util::*;
diff --git a/internal/core/thirdparty/tantivy/tantivy-binding/src/analyzer/filter/synonym_filter.rs b/internal/core/thirdparty/tantivy/tantivy-binding/src/analyzer/filter/synonym_filter.rs
new file mode 100644
index 0000000000..bc50bd5f09
--- /dev/null
+++ b/internal/core/thirdparty/tantivy/tantivy-binding/src/analyzer/filter/synonym_filter.rs
@@ -0,0 +1,364 @@
+use crate::error::{Result, TantivyBindingError};
+use serde_json as json;
+use std::collections::{HashMap, HashSet};
+use std::sync::Arc;
+use tantivy::tokenizer::{Token, TokenFilter, TokenStream, Tokenizer};
+
+pub struct SynonymDictBuilder {
+    dict: HashMap<String, HashSet<String>>,
+    expand: bool,
+}
+
+impl SynonymDictBuilder {
+    fn new(expand: bool) -> SynonymDictBuilder {
+        SynonymDictBuilder {
+            dict: HashMap::new(),
+            expand: expand,
+        }
+    }
+
+    // TODO: Optimize memory usage when add group
+    //       (forbid clone multiple times here)
+    fn add_group(&mut self, words: Vec<String>) {
+        if words.is_empty() {
+            return;
+        }
+
+        for (_, key_word) in words.iter().enumerate() {
+            let push_vec = if self.expand {
+                words.clone()
+            } else {
+                vec![words.first().cloned().unwrap_or_default()]
+            };
+
+            self.add(key_word.clone(), push_vec);
+        }
+    }
+
+    fn add_mapping(&mut self, keys: Vec<String>, words: Vec<String>) {
+        for key in keys {
+            self.add(key, words.clone());
+        }
+    }
+
+    fn add(&mut self, key: String, words: Vec<String>) {
+        if let Some(list) = self.dict.get_mut(&key) {
+            list.extend(words);
+        } else {
+            let mut set: HashSet<_> = words.into_iter().collect();
+            if self.expand {
+                set.insert(key.clone());
+            }
+            self.dict.insert(key, set);
+        }
+    }
+
+    // read row from synonyms dict
+    // use "A, B, C" to represent A, B, C was synonym for each other
+    // use "A => B, C" to represent A will map to B and C
+    // "=>", ",", " " are special characters, should be escaped with "\" if you want to use them as normal characters
+    // synonyms dict don't support space between words, please use "\" to escape space
+    // TODO: synonyms group support space between words
+    fn add_row(&mut self, str: &str) -> Result<()> {
+        let mut is_mapping = false;
+        let mut space_flag = false;
+        let mut left: Vec<String> = Vec::new();
+        let mut right: Vec<String> = Vec::new();
+        let mut current = String::new();
+
+        let chars = str.chars().collect::<Vec<char>>();
+        let mut i = 0;
+        while i < chars.len() {
+            // handle escape
+            if chars[i] == '\\' {
+                if i + 1 >= chars.len() {
+                    return Err(TantivyBindingError::InvalidArgument(format!(
+                        "invalid synonym escaped in the end: {}",
+                        str
+                    )));
+                }
+                if chars[i + 1] == ',' || chars[i + 1] == '\\' || chars[i + 1] == ' ' {
+                    current.push(chars[i + 1]);
+                    i += 2;
+                    continue;
+                }
+
+                if chars[i + 1] == '=' && i + 2 < chars.len() && chars[i + 2] == '>' {
+                    current.push(chars[i + 1]);
+                    current.push(chars[i + 2]);
+                    i += 3;
+                    continue;
+                }
+
+                return Err(TantivyBindingError::InvalidArgument(format!(
+                    "invalid synonym escaped: \\{} in {}",
+                    chars[i + 1],
+                    str,
+                )));
+            }
+
+            // handle space
+            if chars[i] == ' ' {
+                if !current.is_empty() {
+                    // skip space after words and set space flag
+                    while i + 1 < chars.len() && chars[i + 1] == ' ' {
+                        i += 1;
+                    }
+                    space_flag = true;
+                }
+                i += 1;
+                continue;
+            }
+
+            // push current to left or right
+            if chars[i] == ',' {
+                if !current.is_empty() {
+                    if is_mapping {
+                        right.push(current);
+                    } else {
+                        left.push(current);
+                    }
+                }
+                current = String::new();
+                space_flag = false;
+                i += 1;
+                continue;
+            }
+
+            // handle mapping
+            if chars[i] == '=' && i + 1 < chars.len() && chars[i + 1] == '>' {
+                if is_mapping {
+                    return Err(TantivyBindingError::InvalidArgument(format!(
+                        "read synonym dict failed, has more than one \"=>\" in {}",
+                        str,
+                    )));
+                } else {
+                    is_mapping = true;
+                    if !current.is_empty() {
+                        left.push(current);
+                    }
+                    current = String::new();
+                    space_flag = false;
+                }
+                i += 2;
+                continue;
+            }
+
+            if space_flag {
+                return Err(TantivyBindingError::InvalidArgument(format!(
+                    "read synonym dict failed, has space between words {}, please use \\ to escape space",
+                    str,
+                )));
+            }
+
+            current.push(chars[i]);
+            i += 1;
+        }
+
+        // push remaining to left or right
+        if !current.is_empty() {
+            if is_mapping {
+                right.push(current);
+            } else {
+                left.push(current);
+            }
+        }
+
+        // add to dict
+        if is_mapping {
+            self.add_mapping(left, right);
+        } else {
+            self.add_group(left);
+        }
+
+        Ok(())
+    }
+
+    fn build(self) -> SynonymDict {
+        SynonymDict::new(self.dict)
+    }
+}
+
+pub struct SynonymDict {
+    dict: HashMap<String, Box<[String]>>,
+}
+
+impl SynonymDict {
+    fn new(dict: HashMap<String, HashSet<String>>) -> SynonymDict {
+        let mut box_dict: HashMap<String, Box<[String]>> = HashMap::new();
+        for (k, v) in dict {
+            box_dict.insert(k, v.into_iter().collect::<Vec<_>>().into_boxed_slice());
+        }
+        return SynonymDict { dict: box_dict };
+    }
+
+    fn get(&self, k: &str) -> Option<&Box<[String]>> {
+        self.dict.get(k)
+    }
+}
+
+#[derive(Clone)]
+pub struct SynonymFilter {
+    dict: Arc<SynonymDict>,
+}
+
+impl SynonymFilter {
+    pub fn from_json(params: &json::Map<String, json::Value>) -> Result<SynonymFilter> {
+        let expand = params.get("expand").map_or(Ok(true), |v| {
+            v.as_bool().ok_or(TantivyBindingError::InvalidArgument(
+                "create synonym filter failed, `expand` must be bool".to_string(),
+            ))
+        })?;
+
+        let mut builder = SynonymDictBuilder::new(expand);
+        if let Some(dict) = params.get("synonyms") {
+            dict.as_array()
+                .ok_or(TantivyBindingError::InvalidArgument(
+                    "create synonym filter failed, `synonyms` must be array".to_string(),
+                ))?
+                .iter()
+                .try_for_each(|v| {
+                    let s = v.as_str().ok_or(TantivyBindingError::InvalidArgument(
+                        "create synonym filter failed, item in `synonyms` must be string"
+                            .to_string(),
+                    ))?;
+                    builder.add_row(s)
+                })?;
+        }
+
+        Ok(SynonymFilter {
+            dict: Arc::new(builder.build()),
+        })
+    }
+}
+
+pub struct SynonymFilterStream<T> {
+    dict: Arc<SynonymDict>,
+    buffer: Vec<Token>,
+    cursor: usize,
+    tail: T,
+}
+
+impl TokenFilter for SynonymFilter {
+    type Tokenizer<T: Tokenizer> = SynonymFilterWrapper<T>;
+
+    fn transform<T: Tokenizer>(self, tokenizer: T) -> SynonymFilterWrapper<T> {
+        SynonymFilterWrapper {
+            dict: self.dict,
+            inner: tokenizer,
+        }
+    }
+}
+
+#[derive(Clone)]
+pub struct SynonymFilterWrapper<T> {
+    dict: Arc<SynonymDict>,
+    inner: T,
+}
+
+impl<T: Tokenizer> Tokenizer for SynonymFilterWrapper<T> {
+    type TokenStream<'a> = SynonymFilterStream<T::TokenStream<'a>>;
+
+    fn token_stream<'a>(&'a mut self, text: &'a str) -> Self::TokenStream<'a> {
+        SynonymFilterStream {
+            dict: self.dict.clone(),
+            buffer: vec![],
+            cursor: 0,
+            tail: self.inner.token_stream(text),
+        }
+    }
+}
+
+impl<T: TokenStream> SynonymFilterStream<T> {
+    fn buffer_empty(&self) -> bool {
+        return self.cursor >= self.buffer.len();
+    }
+
+    fn next_tail(&mut self) -> bool {
+        if self.tail.advance() {
+            let token = self.tail.token();
+            self.buffer.clear();
+            self.cursor = 0;
+            if let Some(list) = self.dict.get(&token.text) {
+                if list.is_empty() {
+                    return true;
+                }
+
+                for s in list {
+                    self.buffer.push(Token {
+                        offset_from: token.offset_from,
+                        offset_to: token.offset_to,
+                        position: token.position,
+                        text: s.clone(),
+                        position_length: token.position_length,
+                    });
+                }
+            }
+            return true;
+        }
+        false
+    }
+}
+
+impl<T: TokenStream> TokenStream for SynonymFilterStream<T> {
+    fn advance(&mut self) -> bool {
+        if !self.buffer_empty() {
+            self.cursor += 1;
+        }
+
+        if self.buffer_empty() {
+            return self.next_tail();
+        }
+        true
+    }
+
+    fn token(&self) -> &Token {
+        if !self.buffer_empty() {
+            return &self.buffer.get(self.cursor).unwrap();
+        }
+        self.tail.token()
+    }
+
+    fn token_mut(&mut self) -> &mut Token {
+        self.tail.token_mut()
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::SynonymFilter;
+    use crate::analyzer::tokenizers::standard_builder;
+    use crate::log::init_log;
+    use serde_json as json;
+    use std::collections::HashSet;
+
+    #[test]
+    fn test_synonym_filter() {
+        init_log();
+        let params = r#"{
+            "type": "synonym",
+            "expand": false,
+            "synonyms": ["trans => translate, \\=>", "\\\\test, test, tests"]
+        }"#;
+        let json_params = json::from_str::<json::Value>(&params).unwrap();
+        let filter = SynonymFilter::from_json(json_params.as_object().unwrap());
+        assert!(filter.is_ok(), "error: {}", filter.err().unwrap());
+        let builder = standard_builder().filter(filter.unwrap());
+        let mut analyzer = builder.build();
+        let mut stream = analyzer.token_stream("test trans synonym");
+
+        let mut results = Vec::<String>::new();
+        while stream.advance() {
+            let token = stream.token();
+            results.push(token.text.clone());
+        }
+
+        assert_eq!(
+            results
+                .iter()
+                .map(|s| s.as_str())
+                .collect::<HashSet<&str>>(),
+            HashSet::from(["\\test", "translate", "=>", "synonym"])
+        );
+    }
+}