diff --git a/.github/workflows/macos.yml b/.github/workflows/macos.yml
index c00bbcec..6860dcdd 100644
--- a/.github/workflows/macos.yml
+++ b/.github/workflows/macos.yml
@@ -9,22 +9,16 @@ jobs:
     strategy:
       matrix:
         include:
-          - name: "macOS 11 + Xcode 11.7"
-            os: macos-11
+          - name: "macOS 13 + Xcode 15.0"
+            os: macos-13
+            arch: x86_64
             compiler: xcode
-            version: "11.7"
-          - name: "macOS 11 + Xcode 12.2"
-            os: macos-11
+            version: "15.0"
+          - name: "macOS 14 Arm64 + Xcode 15.0"
+            os: macos-14
+            arch: arm64
             compiler: xcode
-            version: "12.4"
-          - name: "macOS 11 + gcc-10"
-            os: macos-11
-            compiler: gcc
-            version: "10"
-          - name: "macOS 11 + gcc-11"
-            os: macos-11
-            compiler: gcc
-            version: "11"
+            version: "15.0"
 
     runs-on: ${{ matrix.os }}
     name: ${{ matrix.name }}
@@ -42,8 +36,8 @@ jobs:
         else
           ls -ls /Applications/
           sudo xcode-select -switch /Applications/Xcode_${{ matrix.version }}.app
-          echo "CC=clang" >> $GITHUB_ENV
-          echo "CXX=clang++" >> $GITHUB_ENV
+          echo "CC=$(brew --prefix llvm@15)/bin/clang" >> $GITHUB_ENV
+          echo "CXX=$(brew --prefix llvm@15)/bin/clang++" >> $GITHUB_ENV
         fi
     - name: Configure Build
       run: mkdir build && cd build && cmake -DCMAKE_BUILD_TYPE=Release -DKIWI_JAVA_BINDING=1 ..
@@ -74,11 +68,17 @@ jobs:
         ./build/kiwi-cli-* -m ./ModelGenerator -e -o test.out kowiki1000.txt
         KIWI_ARCH_TYPE=none ./build/kiwi-cli-* -m ./ModelGenerator -e -o test.out kowiki1000.txt
         KIWI_ARCH_TYPE=balanced ./build/kiwi-cli-* -m ./ModelGenerator -e -o test.out kowiki1000.txt
-        KIWI_ARCH_TYPE=sse2 ./build/kiwi-cli-* -m ./ModelGenerator -e -o test.out kowiki1000.txt
-        KIWI_ARCH_TYPE=sse4_1 ./build/kiwi-cli-* -m ./ModelGenerator -e -o test.out kowiki1000.txt
-        KIWI_ARCH_TYPE=avx2 ./build/kiwi-cli-* -m ./ModelGenerator -e -o test.out kowiki1000.txt
-        KIWI_ARCH_TYPE=avx2 ./build/kiwi-cli-* -m ./ModelGenerator -e -o test.out --sbg kowiki1000.txt
-        KIWI_ARCH_TYPE=avx2 ./build/kiwi-cli-* -m ./ModelGenerator -e -o test.out --typos 6 kowiki1000.txt
+        if [ "${{ matrix.arch }}" = "x86_64" ]; then
+          KIWI_ARCH_TYPE=sse2 ./build/kiwi-cli-* -m ./ModelGenerator -e -o test.out kowiki1000.txt
+          KIWI_ARCH_TYPE=sse4_1 ./build/kiwi-cli-* -m ./ModelGenerator -e -o test.out kowiki1000.txt
+          KIWI_ARCH_TYPE=avx2 ./build/kiwi-cli-* -m ./ModelGenerator -e -o test.out kowiki1000.txt
+          KIWI_ARCH_TYPE=avx2 ./build/kiwi-cli-* -m ./ModelGenerator -e -o test.out --sbg kowiki1000.txt
+          KIWI_ARCH_TYPE=avx2 ./build/kiwi-cli-* -m ./ModelGenerator -e -o test.out --typos 6 kowiki1000.txt
+        else
+          KIWI_ARCH_TYPE=neon ./build/kiwi-cli-* -m ./ModelGenerator -e -o test.out kowiki1000.txt
+          KIWI_ARCH_TYPE=neon ./build/kiwi-cli-* -m ./ModelGenerator -e -o test.out --sbg kowiki1000.txt
+          KIWI_ARCH_TYPE=neon ./build/kiwi-cli-* -m ./ModelGenerator -e -o test.out --typos 6 kowiki1000.txt
+        fi
     - name: Archive binaries
       uses: actions/upload-artifact@v2
       with:
diff --git a/.github/workflows/release.yml b/.github/workflows/release.yml
index 8783b9af..641426e8 100644
--- a/.github/workflows/release.yml
+++ b/.github/workflows/release.yml
@@ -144,7 +144,7 @@ jobs:
   build-macos:
     strategy:
       matrix:
-          os: [macos-11]
+          os: [macos-13]
           arch: [x86_64, arm64]
 
     runs-on: ${{ matrix.os }}
diff --git a/CMakeLists.txt b/CMakeLists.txt
index 51c6dd02..93bb0382 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -1,6 +1,6 @@
 cmake_minimum_required(VERSION 3.12)
 
-project(kiwi VERSION 0.17.1 DESCRIPTION "Kiwi, Korean Intelligent Word Identifier")
+project(kiwi VERSION 0.18.0 DESCRIPTION "Kiwi, Korean Intelligent Word Identifier")
 
 set ( CMAKE_CXX_STANDARD 14 )
 set ( CMAKE_VERBOSE_MAKEFILE true )
@@ -48,7 +48,7 @@ set ( CORE_SRCS
   src/Form.cpp
   src/FeatureTestor.cpp
   src/FileUtils.cpp
-  src/HSDataset.cpp
+  src/Dataset.cpp
   src/Joiner.cpp
   src/Kiwi.cpp
   src/KiwiBuilder.cpp
@@ -56,6 +56,7 @@ set ( CORE_SRCS
   src/PatternMatcher.cpp
   src/search.cpp
   src/ScriptType.cpp
+  src/SubstringExtractor.cpp
   src/SwTokenizer.cpp
   src/TagUtils.cpp
   src/TypoTransformer.cpp
diff --git a/bindings/java/JniUtils.hpp b/bindings/java/JniUtils.hpp
index 10fd70be..f0cb47d9 100644
--- a/bindings/java/JniUtils.hpp
+++ b/bindings/java/JniUtils.hpp
@@ -924,7 +924,8 @@ namespace jni
 			{
 				return CppType{ env, v };
 			}
-			if (!env->IsInstanceOf(v, JIteratorBase::jClass)) throw std::runtime_error{ StringConcat_v<svNotInstanceOf, typeStr, svNullTerm>.data()};
+			// The following line crashes clang compiler. I don't know why, but it's not necessary. So I commented it out.
+			if (!env->IsInstanceOf(v, JIteratorBase::jClass)) throw std::runtime_error{ ""/*StringConcat_v<svNotInstanceOf, typeStr, svNullTerm>.data()*/};
 			return CppType{ env, v };
 		}
 	};
diff --git a/bindings/java/kr/pe/bab2min/Kiwi.java b/bindings/java/kr/pe/bab2min/Kiwi.java
index 0f343779..e8a607de 100644
--- a/bindings/java/kr/pe/bab2min/Kiwi.java
+++ b/bindings/java/kr/pe/bab2min/Kiwi.java
@@ -12,7 +12,7 @@
 
 public class Kiwi implements AutoCloseable  {
 	private long _inst;
-	final private static String _version = "0.17.1";
+	final private static String _version = "0.18.0";
 
 	public static class Match {
 		final static public int none = 0,
diff --git a/include/kiwi/HSDataset.h b/include/kiwi/Dataset.h
similarity index 100%
rename from include/kiwi/HSDataset.h
rename to include/kiwi/Dataset.h
diff --git a/include/kiwi/Form.h b/include/kiwi/Form.h
index 3b8f294e..38eda09a 100644
--- a/include/kiwi/Form.h
+++ b/include/kiwi/Form.h
@@ -2,8 +2,8 @@
  * @file Form.h
  * @author bab2min (bab2min@gmail.com)
  * @brief 형태 및 형태소에 관한 정보를 담는 구조체들이 선언된 헤더
- * @version 0.17.0
- * @date 2022-09-01
+ * @version 0.18.0
+ * @date 2024-07-01
  * 
  * 
  */
diff --git a/include/kiwi/Kiwi.h b/include/kiwi/Kiwi.h
index e052cbaf..f69c4eb7 100644
--- a/include/kiwi/Kiwi.h
+++ b/include/kiwi/Kiwi.h
@@ -2,8 +2,8 @@
  * @file Kiwi.h
  * @author bab2min (bab2min@gmail.com)
  * @brief Kiwi C++ API를 담고 있는 헤더 파일
- * @version 0.17.0
- * @date 2022-09-01
+ * @version 0.18.0
+ * @date 2024-07-01
  * 
  * 
  */
@@ -650,7 +650,7 @@ namespace kiwi
 		 * @param numThreads 모델 및 형태소 분석에 사용할 스레드 개수
 		 * @param options 생성 옵션. `kiwi::BuildOption`을 참조
 		 */
-		KiwiBuilder(const std::string& modelPath, size_t numThreads = 0, BuildOption options = BuildOption::integrateAllomorph | BuildOption::loadDefaultDict, bool useSBG = false);
+		KiwiBuilder(const std::string& modelPath, size_t numThreads = 0, BuildOption options = BuildOption::default_, bool useSBG = false);
 
 		/**
 		 * @brief 현재 KiwiBuilder 객체가 유효한 분석 모델을 로딩한 상태인지 알려준다.
diff --git a/include/kiwi/Macro.h b/include/kiwi/Macro.h
index 25b0c393..bda5ef8b 100644
--- a/include/kiwi/Macro.h
+++ b/include/kiwi/Macro.h
@@ -4,7 +4,7 @@
 #define KIWI_STR(x) KIWI_STR_HELPER(x)
 
 #define KIWI_VERSION_MAJOR 0
-#define KIWI_VERSION_MINOR 17
-#define KIWI_VERSION_PATCH 1
+#define KIWI_VERSION_MINOR 18
+#define KIWI_VERSION_PATCH 0
 
 #define KIWI_VERSION_STRING KIWI_STR(KIWI_VERSION_MAJOR) "." KIWI_STR(KIWI_VERSION_MINOR) "." KIWI_STR(KIWI_VERSION_PATCH)
diff --git a/include/kiwi/SubstringExtractor.h b/include/kiwi/SubstringExtractor.h
new file mode 100644
index 00000000..c4ba7c90
--- /dev/null
+++ b/include/kiwi/SubstringExtractor.h
@@ -0,0 +1,16 @@
+#pragma once
+
+#include <vector>
+#include <string>
+
+namespace kiwi
+{
+	std::vector<std::pair<std::u16string, size_t>> extractSubstrings(
+		const char16_t* first, 
+		const char16_t* last, 
+		size_t minCnt, 
+		size_t minLength = 2, 
+		size_t maxLength = 32,
+		bool longestOnly = true,
+		char16_t stopChr = 0);
+}
diff --git a/include/kiwi/SwTokenizer.h b/include/kiwi/SwTokenizer.h
index 64d61053..4019aa8b 100644
--- a/include/kiwi/SwTokenizer.h
+++ b/include/kiwi/SwTokenizer.h
@@ -2,8 +2,8 @@
  * @file SwTokenizer.h
  * @author bab2min (bab2min@gmail.com)
  * @brief Subword Tokenizer
- * @version 0.16.1
- * @date 2022-07-28
+ * @version 0.18.0
+ * @date 2024-07-01
  *
  *
  */
diff --git a/include/kiwi/Types.h b/include/kiwi/Types.h
index ea1fa317..6af80524 100644
--- a/include/kiwi/Types.h
+++ b/include/kiwi/Types.h
@@ -2,8 +2,8 @@
  * @file Types.h
  * @author bab2min (bab2min@gmail.com)
  * @brief Kiwi C++ API에 쓰이는 주요 타입들을 모아놓은 헤더 파일
- * @version 0.17.0
- * @date 2022-09-01
+ * @version 0.18.0
+ * @date 2024-07-01
  * 
  * 
  */
diff --git a/include/kiwi/capi.h b/include/kiwi/capi.h
index d70a36b5..24591421 100644
--- a/include/kiwi/capi.h
+++ b/include/kiwi/capi.h
@@ -2,8 +2,8 @@
  * @file capi.h
  * @author bab2min (bab2min@gmail.com)
  * @brief Kiwi C API를 담고 있는 헤더 파일
- * @version 0.17.0
- * @date 2022-09-01
+ * @version 0.18.0
+ * @date 2024-07-01
  * 
  * 
  */
@@ -45,7 +45,11 @@ typedef struct {
 	uint32_t line_number; /**< 줄 번호*/
 	uint16_t length; /**< 길이(UTF16 문자 기준) */
 	uint8_t tag; /**< 품사 태그 */
-	uint8_t sense_id; /**< 의미 번호 */
+	union
+	{
+		uint8_t sense_id; /**< 의미 번호 */
+		uint8_t script; /**< 유니코드 영역에 기반한 문자 타입 */
+	};
 	float score; /**< 해당 형태소의 언어모델 점수 */
 	float typo_cost; /**< 오타가 교정된 경우 오타 비용. 그렇지 않은 경우 0 */
 	uint32_t typo_form_id; /**< 교정 전 오타의 형태에 대한 정보 (typoCost가 0인 경우 의미 없음) */
@@ -1008,6 +1012,16 @@ DECL_DLL int kiwi_pt_add_token_to_span_w(kiwi_pretokenized_h handle, int span_id
  */
 DECL_DLL int kiwi_pt_close(kiwi_pretokenized_h handle);
 
+/**
+ * @brief `kiwi_token_info_t`의 `script`가 가리키는 문자 영역의 유니코드 상 이름을 반환합니다.
+ *
+ * @param script `kiwi_token_info_t`의 `script` 필드 값
+ * @return 유니코드 영역의 이름을 반환합니다. 알 수 없을 경우 "Unknown"을 반환합니다.
+ *
+ * @note 이 함수가 반환하는 값은 string literal이므로 별도로 해제할 필요가 없습니다.
+ */
+DECL_DLL const char* kiwi_get_script_name(uint8_t script);
+
 #ifdef __cplusplus  
 }
 #endif 
diff --git a/src/HSDataset.cpp b/src/Dataset.cpp
similarity index 99%
rename from src/HSDataset.cpp
rename to src/Dataset.cpp
index f5f2e741..4a266ec4 100644
--- a/src/HSDataset.cpp
+++ b/src/Dataset.cpp
@@ -1,4 +1,4 @@
-#include <kiwi/HSDataset.h>
+#include <kiwi/Dataset.h>
 #include "RaggedVector.hpp"
 
 using namespace kiwi;
diff --git a/src/KiwiBuilder.cpp b/src/KiwiBuilder.cpp
index e5b2662d..06225241 100644
--- a/src/KiwiBuilder.cpp
+++ b/src/KiwiBuilder.cpp
@@ -3,7 +3,7 @@
 
 #include <kiwi/Kiwi.h>
 #include <kiwi/Utils.h>
-#include <kiwi/HSDataset.h>
+#include <kiwi/Dataset.h>
 #include "ArchAvailable.h"
 #include "KTrie.h"
 #include "StrUtils.h"
diff --git a/src/PatternMatcher.cpp b/src/PatternMatcher.cpp
index ac1e5b39..2eb054c8 100644
--- a/src/PatternMatcher.cpp
+++ b/src/PatternMatcher.cpp
@@ -272,7 +272,7 @@ size_t PatternMatcherImpl::testAbbr(const char16_t* first, const char16_t* last)
 	if (b != last && *b == ' ')
 	{
 		if (l > (isUpperAlpha(*first) ? 5 : 3)) return 0; // reject too long patterns for abbreviation
-		++b;
+		return b - first;
 	}
 	else
 	{
@@ -287,7 +287,6 @@ size_t PatternMatcherImpl::testAbbr(const char16_t* first, const char16_t* last)
 		
 		if (b != last && *b == '.') ++b;
 		else return b - first;
-		if (b != last && *b == ' ') ++b;
 	}
 	if (b[-1] == ' ') --b;
 	return b - first;
diff --git a/src/ScriptType.cpp b/src/ScriptType.cpp
index 7c23ca8a..7328038e 100644
--- a/src/ScriptType.cpp
+++ b/src/ScriptType.cpp
@@ -563,7 +563,7 @@ namespace kiwi
         if (type == ScriptType::chess_symbols) return "Chess Symbols";
         if (type == ScriptType::symbols_for_legacy_computing) return "Symbols for Legacy Computing";
         if (type == ScriptType::tags) return "Tags";
-        return "unknown";
+        return "Unknown";
     }
 
     int isEmoji(char32_t c0, char32_t c1)
diff --git a/src/SubstringExtractor.cpp b/src/SubstringExtractor.cpp
new file mode 100644
index 00000000..bdcaffda
--- /dev/null
+++ b/src/SubstringExtractor.cpp
@@ -0,0 +1,141 @@
+#include <algorithm>
+
+#include <kiwi/Types.h>
+#include <kiwi/SubstringExtractor.h>
+
+#include "StrUtils.h"
+
+#include "sais/fm_index.hpp"
+
+using namespace std;
+
+namespace kiwi
+{
+	inline bool testRepetition(const char16_t* s, size_t l)
+	{
+		if (l < 5) return false;
+		for (size_t i = 1; i <= l / 3; ++i)
+		{
+			bool all = true;
+			for (size_t j = 1; j < l / i; ++j)
+			{
+				if (!equal(s, s + i, s + i * j))
+				{
+					all = false;
+					break;
+				}
+			}
+			if (all) return true;
+		}
+		return false;
+	}
+
+	vector<pair<u16string, size_t>> extractSubstrings(
+		const char16_t* first, 
+		const char16_t* last, 
+		size_t minCnt, 
+		size_t minLength,
+		size_t maxLength,
+		bool longestOnly,
+		char16_t stopChr
+	)
+	{
+		Vector<char16_t> buf(last - first + 1);
+		copy(first, last, buf.begin() + 1);
+		sais::FmIndex<char16_t> fi{ buf.data(), buf.size() };
+		Vector<UnorderedMap<u16string, size_t>> candCnts;
+		vector<pair<u16string, size_t>> ret;
+		fi.enumSuffices(minCnt, [&](const sais::FmIndex<char16_t>::SuffixTy& s, const sais::FmIndex<char16_t>::TraceTy& t)
+		{
+			if (s.size() > maxLength) return false;
+
+			if (find(s.begin(), s.end(), stopChr) != s.end())
+			{
+				return false;
+			}
+
+			if (isLowSurrogate(s.back()) || isHighSurrogate(s.front())) return false;
+
+			if (testRepetition(s.data(), s.size()))
+			{
+				return false;
+			}
+
+			const auto ssLength = s.size();
+			if (ssLength < minLength)
+			{
+				return true;
+			}
+
+			const auto ssCnt = t.back().second - t.back().first;
+			if (ssCnt < minCnt)
+			{
+				return true;
+			}
+
+			if (longestOnly)
+			{
+				if (candCnts.size() <= ssLength - minLength) candCnts.resize(ssLength - minLength + 1);
+				candCnts[ssLength - minLength].emplace(u16string{ s.rbegin(), s.rend() }, ssCnt);
+			}
+			else
+			{
+				ret.emplace_back(u16string{ s.rbegin(), s.rend() }, ssCnt);
+			}
+			return true;
+		});
+
+		if (longestOnly)
+		{
+			for (size_t i = 1; i < candCnts.size(); ++i)
+			{
+				auto& cands = candCnts[i];
+				auto& subCands = candCnts[i - 1];
+				for (auto& p : cands)
+				{
+					auto it = subCands.find(p.first.substr(1));
+					if (it != subCands.end() && it->second == p.second)
+					{
+						subCands.erase(it);
+					}
+
+					it = subCands.find(p.first.substr(0, p.first.size() - 1));
+					if (it != subCands.end() && it->second == p.second)
+					{
+						subCands.erase(it);
+					}
+				}
+			}
+
+			for (auto it = candCnts.rbegin(); it != candCnts.rend(); ++it)
+			{
+				size_t offset = ret.size();
+				for (auto& p : *it)
+				{
+					ret.emplace_back(p.first, p.second);
+				}
+				sort(ret.begin() + offset, ret.end(), [&](const pair<u16string, size_t>& a, const pair<u16string, size_t>& b)
+				{
+					return a.second > b.second;
+				});
+			}
+		}
+		else
+		{
+			sort(ret.begin(), ret.end(), [&](const pair<u16string, size_t>& a, const pair<u16string, size_t>& b)
+			{
+				if (a.first.size() > b.first.size())
+				{
+					return true;
+				}
+				else if (a.first.size() < b.first.size())
+				{
+					return false;
+				}
+				return a.second > b.second;
+			});
+		}
+
+		return ret;
+	}
+}
diff --git a/src/capi/kiwi_c.cpp b/src/capi/kiwi_c.cpp
index 74e760e3..88d0fec8 100644
--- a/src/capi/kiwi_c.cpp
+++ b/src/capi/kiwi_c.cpp
@@ -1434,3 +1434,7 @@ int kiwi_pt_close(kiwi_pretokenized_h handle)
 	}
 }
 
+const char* kiwi_get_script_name(uint8_t script)
+{
+	return getScriptName((ScriptType)script);
+}
diff --git a/test/test_c.cpp b/test/test_c.cpp
index f91c73da..257c750a 100644
--- a/test/test_c.cpp
+++ b/test/test_c.cpp
@@ -389,4 +389,11 @@ TEST(KiwiC, Pretokenized)
 		EXPECT_EQ(kiwi_res_close(o), 0);
 		EXPECT_EQ(kiwi_pt_close(pretokenized), 0);
 	}
-}
\ No newline at end of file
+}
+
+TEST(KiwiC, ScriptName)
+{
+	EXPECT_STREQ(kiwi_get_script_name(0), "Unknown");
+	EXPECT_STREQ(kiwi_get_script_name(1), "Latin");
+	EXPECT_STREQ(kiwi_get_script_name(30), "Hangul");
+}
diff --git a/test/test_cpp.cpp b/test/test_cpp.cpp
index 299f4fe6..50bda190 100644
--- a/test/test_cpp.cpp
+++ b/test/test_cpp.cpp
@@ -1,6 +1,7 @@
 #include "gtest/gtest.h"
 #include <kiwi/Kiwi.h>
-#include <kiwi/HSDataset.h>
+#include <kiwi/Dataset.h>
+#include <kiwi/SubstringExtractor.h>
 #include "common.h"
 
 class TestInitializer
@@ -53,6 +54,36 @@ Kiwi& reuseKiwiInstance()
 	return kiwi;
 }
 
+TEST(KiwiCpp, ExtractSubstrings)
+{
+	const std::u16string s = u"자, 너 오늘 하루 뭐 했니? "
+		"난 오늘 하루가 좀 단순했지. 음 뭐 했는데? "
+		"아침에 수업 받다가 오구 그~ 학생이, 응. 미찌 학생인데, "
+		"음. 되게 귀엽게 생겼다, 음. 되게 웃으면 곰돌이 인형같이 생겼어. "
+		"곰돌이? 응 아니 곰돌이도 아니구, 어쨌든 무슨 인형같이 생겼어, "
+		"펜더곰 그런 거, 왜 이렇게 닮은 사람을 대 봐. "
+		"내가 아는 사람 중에서, 한무 한무? 금붕어잖아? "
+		"맞어 눈도 이렇게 톡 튀어나오구, 어. 한무? 조금 잘 생긴 한무. "
+		"잘 생긴 게 아니라 귀여운 한무. 귀여운 한무? "
+		"어 학원에서 별명도 귀여운 한무였어. "
+		"응. 눈이 똥그래 가지고 그래? 어. 좀 특이한 사람이구나.";
+	auto substrings = extractSubstrings(s.data(), s.data() + s.size(), 2, 2, 32, true, u' ');
+	EXPECT_EQ(substrings.size(), 23);
+
+	const std::u16string t = u"어. 그러다가 갈아타야 되니까 비몽사몽간에 갈아타는 데로 사람 따라서 쓸려가. "
+		"음. 그러다가 요행히 자리를 차지하고 앉어. 음. 그러면은, 어. 예의상 처음부터는 이렇게 신문을 좀 "
+		"보는 척을 해. 그러다가 까무룩 잠이 드는 거야, 까무룩? 근데 내릴 때는 그래두 기가 막히게 내리지? "
+		"내릴 때 딱 한 번. 어, 들켰어? 넘어 간 넘어 간 적이 있었어, 너 어떻게 했어? 돌아왔지 뭐~, 다시 "
+		"타고 얼마나 갔는데? 한 정거장 더 갔을 거야. 아마 선릉인가 삼성인가에, 어. 내려 가지고 삼성인가 "
+		"보다 두 정거장 더 가서, 어. 다시 왔지. 음. 나는 버스 타고 다니잖아? 음. 근데 학원에 갈 때도 버스 "
+		"타고 아니면 여기 교보빌딩 갈 때도 버스 타잖아? 음 짜증나겠구나, 아니야 되게 가깝잖아, 버스 타고 십 "
+		"분, 막히잖아 광화문터널, 안 막혀 별루 그 시간에. 그래 좋겠다. 일곱시 반부터 여덟시부터는 막히는지 "
+		"모르겠는데 어쨌든 일곱시 반부터 내가 빨리 타면 일곱시 사십분 차를 타고 늦게 타면 사십오분 차를 "
+		"타거던, 음. 그때는 그까 앉아 가는 일은 별루 없어. 음. 별루 애들이 거기도 뭐~ 애들이 아현 거기쯤에 "
+		"내리더라, 학생들이 그쪽 학교 있지? 너두 거기 나왔잖아? 음. 이름이 뭐지? 한성! 어이구 아네,";
+	substrings = extractSubstrings(t.data(), t.data() + t.size(), 3, 3, 32, true, u' ');
+	EXPECT_EQ(substrings.size(), 3);
+}
 
 TEST(KiwiCpp, InitClose)
 {
diff --git a/vsproj/kiwi_shared_library.vcxproj b/vsproj/kiwi_shared_library.vcxproj
index c70f01cc..35a21b68 100644
--- a/vsproj/kiwi_shared_library.vcxproj
+++ b/vsproj/kiwi_shared_library.vcxproj
@@ -32,7 +32,7 @@
     <ClInclude Include="..\include\kiwi\FixedVector.hpp" />
     <ClInclude Include="..\include\kiwi\Form.h" />
     <ClInclude Include="..\include\kiwi\FrozenTrie.h" />
-    <ClInclude Include="..\include\kiwi\HSDataset.h" />
+    <ClInclude Include="..\include\kiwi\Dataset.h" />
     <ClInclude Include="..\include\kiwi\Joiner.h" />
     <ClInclude Include="..\include\kiwi\Kiwi.h" />
     <ClInclude Include="..\include\kiwi\Knlm.h" />
@@ -42,6 +42,7 @@
     <ClInclude Include="..\include\kiwi\PatternMatcher.h" />
     <ClInclude Include="..\include\kiwi\ScriptType.h" />
     <ClInclude Include="..\include\kiwi\SkipBigramModel.h" />
+    <ClInclude Include="..\include\kiwi\SubstringExtractor.h" />
     <ClInclude Include="..\include\kiwi\TagUtils.h" />
     <ClInclude Include="..\include\kiwi\TemplateUtils.hpp" />
     <ClInclude Include="..\include\kiwi\ThreadPool.h" />
@@ -115,9 +116,11 @@
     </ClCompile>
     <ClCompile Include="..\src\ArchUtils.cpp" />
     <ClCompile Include="..\src\Combiner.cpp" />
+    <ClCompile Include="..\src\Dataset.cpp" />
     <ClCompile Include="..\src\FileUtils.cpp" />
     <ClCompile Include="..\src\Joiner.cpp" />
     <ClCompile Include="..\src\search.cpp" />
+    <ClCompile Include="..\src\SubstringExtractor.cpp" />
     <ClCompile Include="..\src\TagUtils.cpp" />
     <ClCompile Include="..\src\ScriptType.cpp" />
     <ClCompile Include="..\src\SwTokenizer.cpp" />