"... In PAGE 8: ... One problem with this system is that it is based on an incomplete revelation of preferences: there is no requirement for a voter to rank the options for which he (she) did not vote. As Table2 shows, on the basis of votes cast by 60 voters, A wins by plurality, yet A would lose against B alone (25 to 35) and against C alone (23 to 37). Table 2: Plurality Voting ... In PAGE 8: ...This then points to a second defect of plurality voting which is the fact that it is subject to agenda manipulation and that the presence, or absence, of options - even if those options cannot win - can affect the outcome. The alternative is for each voter to rank the alternatives in order of preference (as in Table2 above) and then the appropriate electoral rule would aggregate these individual rankings into an overall ranking. Such a procedure is termed an apos;ordinal procedure apos;.... In PAGE 8: ...ankings into an overall ranking. Such a procedure is termed an apos;ordinal procedure apos;. One possible electoral rule, based on an ordinal procedure, is the Borda count: in the presence of N options, assign N points to the option ranked first, N-1 points to the option ranked second and finally one point to the option ranked last. A Borda count applied to the data in Table2 sees C a comfortable winner with 138 points, A coming second with 105 points and B finishing last with 91 points. The Borda count method, however, is also susceptible to false revelation of preferences: voters, irrespective of their true preferences, would be inclined to give the lowest preference vote to the candidate they thought was most threatening to their preferred candidates electoral prospects (Miller, 1987).... In PAGE 9: ... An alternative that wins over all the others is then selected the preferred option and is termed the Condorcet winner. Thus, in Table2 , the Condorcet winner C beats A, 37-23 and beats B, 41-19. However, as Table 1 showed, and as Table 4 shows, a Condorcet winner need not exist: in Table 3 demonstrates the phenomenon of apos;cyclical voting apos; - also termed the apos;paradox of voting apos; - whereby A beats B (33-27); B beats C (42-18); and C beats A (35-25).... ..."

"... In PAGE 32: ... Switching between persons will confuse the reader. Table5 provides a list of singular and plural pronouns. Table 5.... ..."

"... In PAGE 3: ...) t followed by the bigram and I t capitalized and followed by a pronoun within 15 tokens Table 2: Feature sets last names are mixed.) In addition, we constructed some composite dictionary features, as specified in Table3 : for example, a word that is in the first-name dictionary and is not in the common-words or last- name dictionaries is designated a sure first name . The common-words dictionary used consists of base forms, conjugations and plural forms of com- mon English words, and a relatively small ad-hoc dictionary representing words especially common in email (e.... In PAGE 4: ...2 Experimental Results We now turn to evaluate the usefulness of the fea- ture sets described above. Table3 gives entity-level F1 performance 5 for CRF trained models for all datasets, using the basic features alone (B); the ba- sic and email-tailored features (B+E); the basic and dictionary features (B+D); and, all of the feature sets combined (B+D+E). All feature sets were tuned us-... ..."

"... In PAGE 7: ... Therefore, we can infer that this might be an example of orthographic interference. It is clear from Table3 that pronouncing the plural /n/ is indeed largely confined to read speech. However, this example also demonstrates the danger of blind trust in global corpus statistics.... ..."