| 1 | = Stylometry = |
| 2 | |
| 3 | [[https://is.muni.cz/auth/predmet/fi/ia161|IA161]] [[en/AdvancedNlpCourse|Advanced NLP Course]], Course Guarantee: Aleš Horák |
| 4 | |
| 5 | Prepared by: Honza Rygl, Aleš Horák |
| 6 | |
| 7 | == State of the Art == |
| 8 | |
| 9 | The analysis of author's characteristic |
| 10 | writing style and vocabulary has been used to uncover author's traits such as authorship, age, or gender |
| 11 | documents by both manual linguistic approaches and automatic algorithmic methods. |
| 12 | |
| 13 | The most common approach to stylometry problems |
| 14 | is to combine stylistic analysis with machine learning techniques: |
| 15 | 1. specific style markers are extracted, |
| 16 | 2. a classification procedure is applied to extracted markers |
| 17 | |
| 18 | |
| 19 | === References === |
| 20 | |
| 21 | 1. Bevendorff, Janek, et al.(2020), Overview of PAN 2020: Authorship Verification, Celebrity Profiling, Profiling Fake News Spreaders on Twitter, and Style Change Detection. International Conference of the Cross-Language Evaluation Forum for European Languages. Springer, Cham. [https://pan.webis.de/downloads/publications/papers/bevendorff_2020.pdf pdf] |
| 22 | 1. Stamatatos, E. (2009), A Survey of Modern Authorship Attribution Methods (2009), Journal of the American Society for Information Science and Technology, 60(3), 538-556. [[http://www.clips.ua.ac.be/~walter/educational/material/Stamatatos_survey2009.pdf | pdf]] |
| 23 | 2. Kestemont, M. (2014), Function Words in Authorship Attribution From Black Magic to Theory? Proceedings of the 3rd Workshop on Computational Linguistics for Literature, EACL 2014, 59–66 [[http://aclweb.org/anthology/W14-0908 | pdf]] |
| 24 | 1. Daelemans, W. (2013). Explanation in computational stylometry. In International conference on intelligent text processing and computational linguistics (pp. 451-462). Springer, Berlin, Heidelberg. [https://www.clips.uantwerpen.be/sites/default/files/daelemans2013.pdf pdf] |
| 25 | |
| 26 | == Practical Session == |
| 27 | |
| 28 | Students will work with the ''Style & Identity Recognition'' (SIR) tool. They will test this tool on prepared data. |
| 29 | The goal will be to implement a small function to extract style markers from a text. |
| 30 | |
| 31 | 1. go to `asteria04.fi.muni.cz` server: |
| 32 | {{{ |
| 33 | ssh asteria04.fi.muni.cz |
| 34 | }}} |
| 35 | 2. Download a [[htdocs:bigdata/stylometry-assignment.zip|ZIP with python packages of the assignment]] |
| 36 | {{{ |
| 37 | wget https://nlp.fi.muni.cz/trac/research/chrome/site/bigdata/stylometry-assignment.zip |
| 38 | }}} |
| 39 | 3. Unzip the downloaded file |
| 40 | {{{ |
| 41 | unzip stylometry-assignment.zip |
| 42 | }}} |
| 43 | 4. Go to the unziped folder |
| 44 | {{{ |
| 45 | cd sir-assignment |
| 46 | }}} |
| 47 | 5. Test the prepared program that analyses data from on-line dating services to distinguish gender (masculine/feminine) by text style features |
| 48 | {{{ |
| 49 | ./run.sh |
| 50 | }}} |
| 51 | |
| 52 | `run.sh` can have two optional parameters: |
| 53 | {{{ |
| 54 | ./run.sh [number_of_testing_cycles] [show_first_N_erroneously_predicted_documents] |
| 55 | }}} |
| 56 | The default values, i.e. running `./run.sh` without parameters, are `10` cycles and `no documents` (`./run.sh 10 0`). With longer feature testing `./run.sh 100` could provide better results (but not necessarily). |
| 57 | |
| 58 | Example with document output (second parameter `>0`): |
| 59 | {{{ |
| 60 | [xrygl@asteria04:~/temp/sir-assignment]$ ./run.sh 10 1 |
| 61 | pos: 5 |
| 62 | expected: on |
| 63 | predicted: ona |
| 64 | text: Ahoj, (nejen) pro výlety do víru podivnězimního velkoměsta, či divočiny |
| 65 | venkova, hledá se partnerka přiměřených rozměrů, tvarů a úrovně. Slečny |
| 66 | veselé povahy preferovány; ona je to nejspíš nutnost :-) |
| 67 | morphology: [ |
| 68 | 1. <s> <s> |
| 69 | 2. Ahoj N.N.I.S.1.-.-.-.-.-.A.-.-.-.- |
| 70 | 3. , Z.:.-.-.-.-.-.-.-.-.-.-.-.-.- |
| 71 | 4. ( Z.:.-.-.-.-.-.-.-.-.-.-.-.-.- |
| 72 | 5. nejen T.T.-.-.-.-.-.-.-.-.-.-.-.-.- |
| 73 | 6. ) Z.:.-.-.-.-.-.-.-.-.-.-.-.-.- |
| 74 | 7. pro R.R.-.-.4.-.-.-.-.-.-.-.-.-.- |
| 75 | 8. výlety N.N.I.P.4.-.-.-.-.-.A.-.-.-.- |
| 76 | 9. do R.R.-.-.2.-.-.-.-.-.-.-.-.-.- |
| 77 | 10. víru N.N.I.S.2.-.-.-.-.-.A.-.-.-.- |
| 78 | 11. podivnězimního A.A.N.S.2.-.-.-.-.1.A.-.-.-.- |
| 79 | 12. velkoměsta N.N.N.S.2.-.-.-.-.-.A.-.-.-.- |
| 80 | 13. , Z.:.-.-.-.-.-.-.-.-.-.-.-.-.- |
| 81 | 14. či J.^.-.-.-.-.-.-.-.-.-.-.-.-.- |
| 82 | 15. divočiny N.N.F.S.2.-.-.-.-.-.A.-.-.-.- |
| 83 | 16. venkova N.N.I.S.2.-.-.-.-.-.A.-.-.-.- |
| 84 | 17. , Z.:.-.-.-.-.-.-.-.-.-.-.-.-.- |
| 85 | 18. hledá V.B.-.S.-.-.-.3.P.-.A.A.-.-.- |
| 86 | 19. se P.7.-.X.4.-.-.-.-.-.-.-.-.-.- |
| 87 | 20. partnerka N.N.F.S.1.-.-.-.-.-.A.-.-.-.- |
| 88 | 21. přiměřených A.A.I.P.2.-.-.-.-.1.A.-.-.-.- |
| 89 | 22. rozměrů N.N.I.P.2.-.-.-.-.-.A.-.-.-.- |
| 90 | 23. , Z.:.-.-.-.-.-.-.-.-.-.-.-.-.- |
| 91 | 24. tvarů N.N.I.P.2.-.-.-.-.-.A.-.-.-.- |
| 92 | 25. a J.^.-.-.-.-.-.-.-.-.-.-.-.-.- |
| 93 | 26. úrovně N.N.F.S.2.-.-.-.-.-.A.-.-.-.- |
| 94 | 27. . Z.:.-.-.-.-.-.-.-.-.-.-.-.-.- |
| 95 | 28. <s> <s> |
| 96 | 29. Slečny N.N.F.P.1.-.-.-.-.-.A.-.-.-.- |
| 97 | 30. veselé A.A.N.S.1.-.-.-.-.1.A.-.-.-.- |
| 98 | 31. povahy N.N.F.S.2.-.-.-.-.-.A.-.-.-.- |
| 99 | 32. preferovány V.s.T.P.-.-.-.X.X.-.A.P.-.-.- |
| 100 | 33. ; Z.:.-.-.-.-.-.-.-.-.-.-.-.-.- |
| 101 | 34. ona P.P.F.S.1.-.-.3.-.-.-.-.-.-.- |
| 102 | 35. je V.B.-.S.-.-.-.3.P.-.A.A.-.-.- |
| 103 | 36. to P.D.N.S.1.-.-.-.-.-.-.-.-.-.- |
| 104 | 37. nejspíš D.g.-.-.-.-.-.-.-.3.A.-.-.-.- |
| 105 | 38. nutnost N.N.F.S.1.-.-.-.-.-.A.-.-.-.- |
| 106 | 39. : Z.:.-.-.-.-.-.-.-.-.-.-.-.-.- |
| 107 | 40. - Z.:.-.-.-.-.-.-.-.-.-.-.-.-.- |
| 108 | 41. ) Z.:.-.-.-.-.-.-.-.-.-.-.-.-.- |
| 109 | ] |
| 110 | Acc: 73.8 +- 2.1% (baseline 50.0%, 10 iterations) |
| 111 | }}} |
| 112 | You may print more details in `http_server/basic_task.py` after `# print explanation(s)` comment. |
| 113 | |
| 114 | === Task === |
| 115 | Examine files in `stylometry_features` folder. |
| 116 | Modify the `assignment.py` file to increase the accuracy of methods (use `vi` or `nano` command for editing). |
| 117 | You can create another classes inside, change their names and class names to improve the accuracy score. Don't forget to add new classes into the `assignments` list in this file. |
| 118 | |
| 119 | Suggestions for your inspiration include: |
| 120 | * diacritics usage (yes/no), a regular expression will be needed |
| 121 | * sentence endings (number of sentences, or typical endings) |
| 122 | * repetitions of words in sentences/in the text |
| 123 | * usage of uppercase letters |
| 124 | * length of sentences/text |
| 125 | * POS tags (n-grams) |
| 126 | * word n-grams |
| 127 | * character n-grams |
| 128 | |
| 129 | Each modification can be tested by running `./run.sh` again. |
| 130 | The first call of `run.sh` can be slower, because documents are morphologically analysed during the first run. |
| 131 | |
| 132 | Write the resulting `Acc:` line into the top comment of the `assignment.py` file. [[br]] |
| 133 | Submit your `assignment.py` file. |