Download at in Visual C#.NET

Printer QR Code in Visual C#.NET Download at

The reason for the .to_f s is so that the lengths are treated as floating decimal point numbers, and the percentage is worked out more accurately. When you work it up to the real percentage (out of 100), you can convert back to an integer once again. You ll see how this all comes together in the final version at the end of this chapter.

Word processors such as Microsoft Word generally have summarization features that can take a long piece of text and seemingly pick out the best sentences to produce an at-a-glance summary. The mechanisms for producing summaries have become more complex over the years, but one of the simplest ways to develop a summarizer of your own is to scan for sentences with certain characteristics. One technique is to look for sentences that are of about average length and that look like they contain nouns. Tiny sentences are unlikely to contain anything useful, and long sentences are likely to be simply too long for a summary. Finding nouns reliably would require systems that are far beyond the scope of this book, so you could cheat by looking for words that indicate the presence of useful nouns in the same sentence, such as is and are (for example, Noun is, Nouns are, There are x nouns ). Let s assume that you want to throw away two-thirds of the sentences a third that are the shortest sentences and a third that are the longest sentences leaving you with an ideal third of the original sentences that are ideally sized for your task. For ease of development, let s create a new program from scratch, and transfer your logic over to the main application later. Create a new program called summarize.rb and use this code: text = %q{ Ruby is a great programming language. It is object oriented and has many groovy features. Some people don't like it, but that's not our problem! It's easy to learn. It's great. To learn more about Ruby, visit the official Ruby Web site today. } sentences = text.gsub(/\s+/, ' ').strip.split(/\.|\ |!/) sentences_sorted = sentences.sort_by { |sentence| sentence.length } one_third = sentences_sorted.length / 3 ideal_sentences = sentences_sorted.slice(one_third, one_third + 1) ideal_sentences = ideal_sentences.select { |sentence| sentence =~ /is|are/ } puts ideal_sentences.join(". ") And for good measure, run it to see what happens: Ruby is a great programming language. It is object oriented and has many groovy features Seems like a success! Let s walk through the program.

First you define the variable text to hold the long string of multiple sentences, much like in analyzer.rb. Next you split text into an array of sentences like so: sentences = text.gsub(/\s+/, ' ').strip.split(/\.|\ |!/) This is slightly different from the method used in analyzer.rb. There is an extra gsub in the chain, as well as strip. The gsub gets rid of all large areas of whitespace and replaces them with a single space (\s+ meaning one or more whitespace characters ). This is simply for cosmetic reasons. The strip removes all extra whitespace from the start and end of the string. The split is then the same as that used in the analyzer. Next you sort the sentences by their lengths, as you want to ignore the shortest third and the longest third: sentences_sorted = sentences.sort_by { |sentence| sentence.length } Arrays and hashes have the sort_by method, which can rearrange them into almost any order you want. sort_by takes a code block as its argument, where the code block is an expression that defines what to sort by. In this case, you re sorting the sentences array. You pass each sentence in as the sentence variable, and choose to sort them by their length, using the length method upon the sentence. After this line, sentences_sorted contains an array with the sentences in length order. Next you need to get the middle third of the length-sorted sentences in sentences_sorted, as these are the ones you ve deemed to be probably the most interesting. To do this, you can divide the length of the array by 3 to get the number of elements in a third, and then grab that number of elements from one third into the array (note that you grab one extra element to compensate for rounding caused by integer division). This is done like so: one_third = sentences_sorted.length / 3 ideal_sentences = sentences_sorted.slice(one_third, one_third + 1) The first line takes the length of the array and divides it by 3 to get the quantity that is equal to a third of the array. The second line uses the slice method to cut out a section of the array to assign to ideal_sentences. In this case, assume that the sentences_sorted is 6 elements long. 6 divided by 3 is 2, so a third of the array is 2 elements long. The slice method then cuts from element 2 for 2 (plus 1) elements, so you effectively carve out elements 2, 3, and 4 (remember that array elements start counting from 0). This means you get the inner third of the ideal-length sentences you wanted. The penultimate line checks to see if the sentence includes the word is or are, and only accepts each sentence if so: ideal_sentences = ideal_sentences.select { |sentence| sentence =~ /is|are/ } It uses the select method, as the stop-word removal code in the previous section did. The expression in the code block uses a regular expression that matches against sentence, and only returns true if is or are are present within sentence. This means ideal_sentences now only contains sentences that are in the middle third lengthwise and contain either is or are. The final line simply joins the ideal_sentences together with a full stop and space between them to make them readable: puts ideal_sentences.join(". ")