diff --git a/DESIGN b/DESIGN
index b6ce4261..7881a81d 100644
--- a/DESIGN
+++ b/DESIGN
@@ -66,40 +66,6 @@ Callbacks:
 
 
 
-Filter:
-	Filtering is used to generate a subset of songs displayed by the UI to
-	that users can choose from.  The inverted index is generated at startup
-	so there is no need for a remove() function, since it will be wiped
-	the next time the application starts.
-
-- Parsing:
-	- Scan over the input text to create a list of words using the following
-	  characters as delimiters: \/,;()_-~+"
-	- While scanning, convert the string to lowercase and strip out any
-	  other special characters.
-
-- API:
-	std::string filter :: add(const std::string &key, unsigned int track_id);
-		Parse the key into words following the "Parsing" section above.
-		Generate substrings for each word and add each (substring,
-		track_id) pair to the index.  Return the lowercased text to the
-		caller.
-
-		To generate substrings, iterate over the word starting from
-		the front.  For example: "goron" would contain the substrings
-		{g, go, gor, goro, goron}.
-
-	std::string filter :: lowercase(const std::string &text);
-		Parse the text into lowercased words following the "Parsing"
-		section above.  Return the lowercased string to the caller.
-
-	void filter :: search(const std::string &text, std::set<track_id> &res);
-		This function finds all track_ids matching the input text.
-		Parse the string into substrings and take the intersection of
-		all sets returned by the index for each substring.
-
-
-
 Idle queue:
 	The idle queue is used to schedule function calls that run at a later
 	time.
diff --git a/TODO b/TODO
index 1471b0c7..7c07132c 100644
--- a/TODO
+++ b/TODO
@@ -86,3 +86,7 @@ Future work:
 		- Show gcov graph
 		- Run cppcheck
 	- Read tags with gstreamer?
+	- Move search functions from filter to index code
+		- Move tag index into tagdb
+		- Rename filter.cpp -> text.cpp (or something to reflect
+		  that it is used for text processing)
diff --git a/core/filter.cpp b/core/filter.cpp
index 8a4edc4a..37689362 100644
--- a/core/filter.cpp
+++ b/core/filter.cpp
@@ -58,12 +58,12 @@ static void parse_text(const std::string &text, std::list<std::string> &ret)
 		ret.push_back(word);
 }
 
-static void add_substrings(const std::string &text, unsigned int track_id)
+static void add_substrings(const std::string &text, unsigned int index)
 {
 	std::string substr;
 	for (unsigned int i = 1; i <= text.size(); i++) {
 		substr = text.substr(0, i);
-		filter_index.insert(substr, track_id);
+		filter_index.insert(substr, index);
 	}
 }
 
@@ -81,14 +81,14 @@ static std::string reassemble_text(std::list<std::string> text)
 	return res;
 }
 
-std::string filter :: add(const std::string &text, unsigned int track_id)
+std::string filter :: add(const std::string &text, unsigned int index)
 {
 	std::list<std::string> parsed;
 	std::list<std::string>::iterator it;
 
 	parse_text(text, parsed);
 	for (it = parsed.begin(); it != parsed.end(); it++)
-		add_substrings(*it, track_id);
+		add_substrings(*it, index);
 	return reassemble_text(parsed);
 }
 
diff --git a/include/core/filter.h b/include/core/filter.h
index 8fdc0b47..3a195758 100644
--- a/include/core/filter.h
+++ b/include/core/filter.h
@@ -9,31 +9,50 @@
 #include <string>
 
 /**
- * Namespace for searching the tag database.
+ * The filter layer is used to find a subset of songs based on an input
+ * string.  Since this layer does text processing, it also provides a
+ * functions for converting strings to lowercase.
+ *
+ * The text processing is mostly interested in alphanumeric characters, so
+ * any special characters included in the input text will be stripped out.
+ * Tabs, spaces and the following characters are used to delimit words:
+ *
+ * 	\/,;()_-~+"
  */
 namespace filter {
 
 	/**
-	 * Break a string into words and add each word to an index.
+	 * Convert the key to lowercase and break into individual words.
+	 * Generate substrings for each word and add each (substring, index)
+	 * pair to the Index.
 	 *
-	 * @param  text  The text to parse.
-	 * @param  id    An identifier paired with the text.
+	 * To generate substrings: iterate over the word starting from the
+	 * first character, and append a character for every iteration.  For
+	 * example, the word "goron" would contain the substrings:
+	 * 	{ g, go, gor, goro, goron }.
+	 *
+	 * @param  text   The text to parse.
+	 * @param  index  The track index to pair with the text.
 	 * @return The lowercase form of the input text.
 	 */
 	std::string add(const std::string &, unsigned int);
 
 	/**
-	 * Search the entire index for identifiers paired with the input text.
+	 * Break the input text into lowercase words and search the Index
+	 * for matches.  The results set should be filled out with the
+	 * intersection of the IndexEntry for each word.  If any word does
+	 * not appear in the index, then the result set should be empty.
 	 *
 	 * @param  text  The text to search for.
-	 * @param  res   The results set will be filled in with the ids found.
+	 * @param  res   The results set to fill in with matching indexes.
 	 */
 	void search(const std::string &, std::set<unsigned int> &);
 
 	/**
-	 * Returns the lowercase version of the input text.
+	 * Converts the input text to lowercase and returns the result.
 	 *
-	 * @param  text  The text to be converted into lowercase.
+	 * @param  text  The text to be converted.
+	 * @return The lowercase form of the input text.
 	 */
 	std::string lowercase(const std::string &);