Experiments

Table: Statistics and example sentences from each domain. s/d = sentences per document; w/d = word tokens per document. The fourth column lists (1) the original document, (2) the document recovered from a count BOW, (3) the document recovered from an indicator BOW, and (4) the document recovered from a stopwords-removed count BOW.

Domain	s/d	w/d	Example
Medical	8	153	She also had some breathlessness .
			She also had some breathlessness .
			She also had some breathlessness .
			We have had some breathlessness .
			He was not wearing a helmet and was seen unconscious when paramedics arrived .
			He was unconscious when paramedics arrived and was seen not wearing a helmet .
			He was not wearing a helmet and was unconscious when paramedics arrived . seen .
			He was seen wearing a helmet was not unconscious when paramedics arrived .
CIA	22	487	These regiments are under a divisional headquarters called the 324 B Division .
			These are called the B 324 Division regiments under a divisional headquarters .
			B Division 324 These are called the regiments under a divisional headquarters .
			324 regiments under Division B of A are called the divisional headquarters .
			The President had his breakfast during the meeting is [sic] the Situation
			Room Conference Room .
			The Situation Room is the President had his breakfast meeting during the
			Conference Room .
			The President had his breakfast . The Situation Room is the meeting
			during the Conference .
			It is during this meeting that the President had breakfast Situation Room
			Conference Room .
Email	11	228	I sincerely wish all of you the best in your future endeavors .
			I sincerely wish all of you the best in your future endeavors .
			I sincerely wish all of you the best of the best in all your future endeavors .
			sincerely wish all the best in future endeavors .
			PIRA is coming in May to do their semi - annual energy outlook .
			May PIRA is coming in to do their semi - annual energy outlook .
			May . PIRA is coming in to do their semi - annual energy outlook .
			PIRA is a semi - annual energy outlook for this coming May .
Stock	13	400	Our quality systems are ISO/TS16949 ( 2002 version ) certified .
			Our quality systems are certified ISO/TS16949 ( 2002 version ) .
			ISO/TS16949 Our quality systems are certified version ) ( 2002 ) .
			ISO/TS16949 quality systems are certified to the 2002 and version ( ) .
			This department operates under the name of Stock Yards Trust Company .
			This department operates under the name of Trust Stock Yards Company .
			This department operates under the name of the Trust . Stock Yards Company .
			Trust in the Stock Yards Company operates under the name department .
SW	111	1771	B : uh - huh um - hum
			B : huh uh - um - hum .
			B : uh huh um - hum .
			Go to : A - B - um uh huh hum .
			A : halfway there so that 's good .
			A halfway there : so that 's good .
			A : halfway there so that 's good .
			so good : it 's halfway there .