diff --git a/cpp/src/arrow/vendored/double-conversion/double-conversion.cc b/cpp/src/arrow/vendored/double-conversion/double-conversion.cc
index d287b44c5d439..8defb333cb9a5 100644
--- a/cpp/src/arrow/vendored/double-conversion/double-conversion.cc
+++ b/cpp/src/arrow/vendored/double-conversion/double-conversion.cc
@@ -85,6 +85,15 @@ void DoubleToStringConverter::CreateExponentialRepresentation(
   ASSERT(length != 0);
   result_builder->AddCharacter(decimal_digits[0]);
 
+  /* If the mantissa of the scientific notation representation is an integer number,
+   * the EMIT_TRAILING_DECIMAL_POINT flag will add a '.' character at the end of the
+   * representation:
+   * - With EMIT_TRAILING_DECIMAL_POINT enabled -> 0.0009 => 9.E-4
+   * - With EMIT_TRAILING_DECIMAL_POINT disabled -> 0.0009 => 9E-4
+   *
+   * If the mantissa is an integer and the EMIT_TRAILING_ZERO_AFTER_POINT flag is enabled
+   * it will add a '0' character at the end of the mantissa representation. Note that that
+   * flag depends on EMIT_TRAILING_DECIMAL_POINT flag be enabled.*/
   if(length == 1){
     if ((flags_ & EMIT_TRAILING_DECIMAL_POINT) != 0) {
       result_builder->AddCharacter('.');
diff --git a/cpp/src/arrow/vendored/double-conversion/double-conversion.h b/cpp/src/arrow/vendored/double-conversion/double-conversion.h
index 6dbc0997c617f..9dc3ebd8dfda3 100644
--- a/cpp/src/arrow/vendored/double-conversion/double-conversion.h
+++ b/cpp/src/arrow/vendored/double-conversion/double-conversion.h
@@ -104,6 +104,17 @@ class DoubleToStringConverter {
   //   ToPrecision(230.0, 2) -> "230"
   //   ToPrecision(230.0, 2) -> "230."  with EMIT_TRAILING_DECIMAL_POINT.
   //   ToPrecision(230.0, 2) -> "2.3e2" with EMIT_TRAILING_ZERO_AFTER_POINT.
+  //
+  // When converting numbers to scientific notation representation, if the mantissa of
+  // the representation is an integer number, the EMIT_TRAILING_DECIMAL_POINT flag will
+  // add a '.' character at the end of the representation:
+  // - With EMIT_TRAILING_DECIMAL_POINT enabled -> 0.0009 => 9.E-4
+  // - With EMIT_TRAILING_DECIMAL_POINT disabled -> 0.0009 => 9E-4
+  //
+  // If the mantissa is an integer and the EMIT_TRAILING_ZERO_AFTER_POINT flag is enabled
+  // it will add a '0' character at the end of the mantissa representation. Note that that
+  // flag depends on EMIT_TRAILING_DECIMAL_POINT flag be enabled.
+  // - With EMIT_TRAILING_ZERO_AFTER_POINT enabled -> 0.0009 => 9.0E-4
   DoubleToStringConverter(int flags,
                           const char* infinity_symbol,
                           const char* nan_symbol,
diff --git a/cpp/src/gandiva/formatting_utils.h b/cpp/src/gandiva/formatting_utils.h
index 257d3077d494a..0f9bef68a7e30 100644
--- a/cpp/src/gandiva/formatting_utils.h
+++ b/cpp/src/gandiva/formatting_utils.h
@@ -39,11 +39,21 @@ class FloatToStringGdvMixin
   using arrow::internal::FloatToStringFormatterMixin<
       ARROW_TYPE>::FloatToStringFormatterMixin;
 
+  // The mixin is a modified version of the existent FloatToStringFormatterMixin, but
+  // it defines some specific parameters in the FloatToStringFormatterMixin to cast
+  // the float numbers to string using the same patterns like Java.
+  //
+  // The Java real numbers are represented in two ways following these rules:
+  //- If the number is greater or equals than 10^7 and less than 10^(-3)
+  //  it will be represented using scientific notation, e.g:
+  //      - 0.000012 -> 1.2E-5
+  //      - 10000002.3 -> 1.00000023E7
+  //- If the numbers are between that interval above, they are showed as is.
   explicit FloatToStringGdvMixin(const std::shared_ptr<arrow::DataType>& = NULLPTR)
       : arrow::internal::FloatToStringFormatterMixin<ARROW_TYPE>(
             DoubleToStringConverter::EMIT_TRAILING_ZERO_AFTER_POINT |
                 DoubleToStringConverter::EMIT_TRAILING_DECIMAL_POINT,
-            "inf", "nan", 'E', -3, 7, 3, 1) {}
+            "Infinity", "NaN", 'E', -3, 7, 3, 1) {}
 };
 
 template <>