This is the sense tagged "line" data, as used in numerous comparative 
studies of word sense disambiguation methodologies. In brief, each 
instance of line has been tagged with one of six WordNet senses. Further
details can be found in the papers mentioned below. 

Unix wc output:

     373   17025   85709 cord2
     376   19559  115350 division2
     349   18196  105154 formation2
     429   21472  124705 phone2
    2218  117598  710036 product2
     404   21473  124368 text2                                                 

This data was first described in the following:

@inproceedings{LeacockTV93,
        author = {Leacock, C. and Towell, G. and Voorhees, E.},
        title = {Corpus-Based Statistical Sense Resolution},
        booktitle = {Proceedings of the ARPA Workshop on Human
                  Language Technology},
        month = {March},
        pages = {260--265},
        year = {1993}}       

(Please credit Leacock et. al. with the creation of this data.  I am
simply distributing this data and played no role in its creation.)

It has since been used and described in the following:

@inproceedings{Pedersen00b,
        author = {Pedersen, T.},
        title = {A Simple Approach to Building Ensembles of Naive Bayesian
Classifiers for Word Sense Disambiguation},
        booktitle = {Proceedings of the North American Chapter of the
Association for Computational Linguistics},
        year = {2000},
        month ={May},
        address = {Seattle, WA}}
                                  
@inproceedings{Pedersen98B,
        author = {Pedersen, T. and Bruce, R.},
        title = {Knowledge Lean Word Sense Disambiguation},
        year = {1998},
        booktitle= {Proceedings of the Fifteenth National Conference on
                  Artificial Intelligence},
        pages = {800--805},
        month = {July},
        address = {Madison, WI}}
                                  
@article{LeacockCM98,
        author = {Leacock, C. and Chodorow, M. and Miller, G.},
        title = {Using Corpus Statistics and {W}ord{N}et Relations for
Sense Identification},
        journal = {Computational Linguistics},
        month = {March},
        volume = {24},
        number = {1},
        pages = {147--165},
        year = {1998}}
                                  
@inproceedings{PedersenB97C,
        author = {Pedersen, T. and Bruce, R.},
        title = {Distinguishing Word Senses in Untagged Text},
        booktitle = {Proceedings of the Second Conference on Empirical
                Methods in Natural Language Processing},
        month = {August},
        year = {1997},
        pages = {197--207},
        address = {Providence, RI}}       

@inproceedings{PedersenB97A,
        author = {Pedersen, T. and Bruce, R.},
        title = {A New Supervised Learning Algorithm for Word Sense
Disambiguation},
        year = {1997},
        booktitle= {Proceedings of the Fourteenth National Conference on
                  Artificial Intelligence},
        month = {July},
        pages = {604--609},
        address = {Providence, RI}}   

@inproceedings{Mooney96,
        author = {Mooney, R.},
        title = {Comparative Experiments on Disambiguating Word
                  Senses: An Illustration of the Role of Bias in
                  Machine Learning },
        booktitle={Proceedings of the Conference on Empirical Methods
                  in Natural Language Processing},
        month = {May},
        pages = {82--91},
        year = {1996}}