I 
I 
I 
I 
i 
I 
| 
i 
i 
i, 
I 
i 
I 
I 
i 
I 
Summarising: Where are we 
should we go? 
Karen Sparek Jones. 
now? Where 
Computer Laboratory, Umverslty of Cambridge 
New Museums Site, Pembroke Street 
Cambmdge CB2 3QG, England 
Karen Sparck-Jones@cl cam ac uk 
Abstract 
Summansmg covers a range from text extraction to content condensation Its 
essential features are picking important concepts from, and reducing, source 
text or mformahon, to dehver summary reformation or text General strategies 
for dolug this are clearly preferable to appheatlon-speclfic ones So far, we have 
found that statlstlcally-baeed sentence extrachon and coneatenatton does not 
produce effective summaries But we have not yet found general methods of 
content analysis and condensation We can only identify key source content 
and present It m summary with heavy domain and goal gmdance The most 
pressing need is to develop 'sufficmnt to the day' techniques that do more than 
surface sentence extraction without depending, MUC-hke, on prior specifica- 
tions for sought content These needed mterme&ate techniques include passage 
extraction and linking, deep phrase selection and ordenng, entity identification 
and relating Such strategies benefit from, or require, shallow text analysis and 
do or can exploit statistical data. They may be enhanced by modern display 
resources They are apphcable to individual source texts or to data sets as 
wholes Most importantly, we can tackle tins level of summarising because cur- 
rent robust parsing technology may succeed, given source redundancy, m getting 
enough of value from sources to help users, and because current text produc- 
tion methods can deliver usable summary texts We should push this line hard, 
seehng to rmmrarse appheatlon-spec~c domain knowledge, to take advantage 
of discourse structure, and to address summary function for the user 
