| 283 |      1 | // Preliminary Part about Code Similarity
 | 
|  |      2 | //========================================
 | 
| 211 |      3 | 
 | 
| 323 |      4 | object CW7a {
 | 
| 320 |      5 | 
 | 
| 211 |      6 | 
 | 
|  |      7 | //(1) Complete the clean function below. It should find
 | 
|  |      8 | //    all words in a string using the regular expression
 | 
| 323 |      9 | //    \w+  and the library function
 | 
| 211 |     10 | //
 | 
|  |     11 | //         some_regex.findAllIn(some_string)
 | 
|  |     12 | //
 | 
|  |     13 | //    The words should be Returned as a list of strings.
 | 
|  |     14 | 
 | 
| 320 |     15 | 
 | 
| 323 |     16 | //def clean(s: String) : List[String] = ...
 | 
|  |     17 | def clean(s: String) : List[String] =
 | 
|  |     18 |     "\\w+".r.findAllIn(s).toList
 | 
| 211 |     19 | 
 | 
| 323 |     20 | //(2) The function occurrences calculates the number of times
 | 
|  |     21 | //    strings occur in a list of strings. These occurrences should
 | 
| 211 |     22 | //    be calculated as a Map from strings to integers.
 | 
|  |     23 | 
 | 
| 320 |     24 | 
 | 
| 323 |     25 | //def occurrences(xs: List[String]): Map[String, Int] = ..
 | 
|  |     26 | def occurrences(xs: List[String]) : Map[String, Int] =
 | 
|  |     27 |     xs.groupBy(identity).view.mapValues(_.size).toMap
 | 
| 211 |     28 | 
 | 
|  |     29 | //(3) This functions calculates the dot-product of two documents
 | 
|  |     30 | //    (list of strings). For this it calculates the occurrence
 | 
| 323 |     31 | //    maps from (2) and then multiplies the corresponding occurrences.
 | 
| 211 |     32 | //    If a string does not occur in a document, the product is zero.
 | 
| 323 |     33 | //    The function finally sums up all products.
 | 
| 211 |     34 | 
 | 
| 320 |     35 | 
 | 
| 323 |     36 | //def prod(lst1: List[String], lst2: List[String]) : Int = ..
 | 
|  |     37 | def prod(lst1: List[String], lst2: List[String]) : Int =
 | 
|  |     38 |     occurrences(lst1).map(x => occurrences(lst2).getOrElse(x._1, 0) * x._2).reduce(_ + _)
 | 
| 320 |     39 | 
 | 
| 211 |     40 | //(4) Complete the functions overlap and similarity. The overlap of
 | 
|  |     41 | //    two documents is calculated by the formula given in the assignment
 | 
|  |     42 | //    description. The similarity of two strings is given by the overlap
 | 
| 323 |     43 | //    of the cleaned strings (see (1)).
 | 
| 211 |     44 | 
 | 
|  |     45 | 
 | 
| 320 |     46 | //def overlap(lst1: List[String], lst2: List[String]) : Double = ...
 | 
| 323 |     47 | def overlap(lst1: List[String], lst2: List[String]) : Double =
 | 
|  |     48 |     prod(lst1, lst2).toDouble/Math.max(prod(lst1, lst1).toDouble, prod(lst2, lst2).toDouble)
 | 
| 320 |     49 | //def similarity(s1: String, s2: String) : Double = ...
 | 
| 323 |     50 | def similarity(s1: String, s2: String) : Double =
 | 
|  |     51 |     overlap(clean(s1), clean(s2))
 | 
| 320 |     52 | 
 | 
|  |     53 | 
 | 
|  |     54 | /* Test cases
 | 
| 323 |     55 | import CW7a._
 | 
|  |     56 | val list1 = List("a", "b", "b", "c", "d")
 | 
| 211 |     57 | val list2 = List("d", "b", "d", "b", "d")
 | 
| 323 |     58 | occurrences(List("a", "b", "b", "c", "d"))
 | 
|  |     59 | occurrences(List("d", "b", "d", "b", "d"))
 | 
|  |     60 | prod(list1,list2) // 7
 | 
| 211 |     61 | overlap(list1, list2)   // 0.5384615384615384
 | 
|  |     62 | overlap(list2, list1)   // 0.5384615384615384
 | 
|  |     63 | overlap(list1, list1)   // 1.0
 | 
|  |     64 | overlap(list2, list2)   // 1.0
 | 
| 323 |     65 | // Plagiarism examples from
 | 
| 211 |     66 | // https://desales.libguides.com/avoidingplagiarism/examples
 | 
|  |     67 | val orig1 = """There is a strong market demand for eco-tourism in
 | 
|  |     68 | Australia. Its rich and diverse natural heritage ensures Australia's
 | 
|  |     69 | capacity to attract international ecotourists and gives Australia a
 | 
|  |     70 | comparative advantage in the highly competitive tourism industry."""
 | 
|  |     71 | val plag1 = """There is a high market demand for eco-tourism in
 | 
|  |     72 | Australia. Australia has a comparative advantage in the highly
 | 
|  |     73 | competitive tourism industry due to its rich and varied natural
 | 
|  |     74 | heritage which ensures Australia's capacity to attract international
 | 
|  |     75 | ecotourists."""
 | 
| 320 |     76 | similarity(orig1, plag1) // 0.8679245283018868
 | 
| 323 |     77 | // Plagiarism examples from
 | 
| 211 |     78 | // https://www.utc.edu/library/help/tutorials/plagiarism/examples-of-plagiarism.php
 | 
|  |     79 | val orig2 = """No oil spill is entirely benign. Depending on timing and
 | 
|  |     80 | location, even a relatively minor spill can cause significant harm to
 | 
|  |     81 | individual organisms and entire populations. Oil spills can cause
 | 
|  |     82 | impacts over a range of time scales, from days to years, or even
 | 
|  |     83 | decades for certain spills. Impacts are typically divided into acute
 | 
|  |     84 | (short-term) and chronic (long-term) effects. Both types are part of a
 | 
|  |     85 | complicated and often controversial equation that is addressed after
 | 
|  |     86 | an oil spill: ecosystem recovery."""
 | 
|  |     87 | val plag2 = """There is no such thing as a "good" oil spill. If the
 | 
|  |     88 | time and place are just right, even a small oil spill can cause damage
 | 
|  |     89 | to sensitive ecosystems. Further, spills can cause harm days, months,
 | 
|  |     90 | years, or even decades after they occur. Because of this, spills are
 | 
|  |     91 | usually broken into short-term (acute) and long-term (chronic)
 | 
|  |     92 | effects. Both of these types of harm must be addressed in ecosystem
 | 
|  |     93 | recovery: a controversial tactic that is often implemented immediately
 | 
|  |     94 | following an oil spill."""
 | 
| 320 |     95 | overlap(clean(orig2), clean(plag2))  // 0.728
 | 
|  |     96 | similarity(orig2, plag2)             // 0.728
 | 
| 323 |     97 | // The punchline: everything above 0.6 looks suspicious and
 | 
| 320 |     98 | // should be investigated by staff.
 | 
| 211 |     99 | */
 | 
|  |    100 | 
 | 
| 320 |    101 | }
 |