Bu makale kapsam\u0131nda ise .NET Core<\/strong><\/em> ve Azure Text Analytics API<\/strong><\/em> kullanarak, bir e-ticaret firmas\u0131ndaki \u00fcr\u00fcn yorumlar\u0131 \u00fczerinde sentiment<\/em><\/strong> analizi ger\u00e7ekle\u015ftirece\u011fiz. Sentiment analizindeki amac\u0131m\u0131z ise bir son kullan\u0131c\u0131n\u0131n bir \u00fcr\u00fcn hakk\u0131ndaki yorumlar\u0131 okumadan, h\u0131zl\u0131 bir \u015fekilde fikir sahibi olabilmesini sa\u011flamak y\u00f6n\u00fcnde olacakt\u0131r.<\/p>\n \u00d6ncelikle sentiment<\/em> analizi nedir, k\u0131saca bir giri\u015f yapmak istiyorum.<\/p>\n K\u0131saca sentiment<\/em> analizi i\u00e7in:<\/p>\n Bir yaz\u0131 par\u00e7as\u0131ndan olumlu<\/em> veya olumsuz<\/em> g\u00f6r\u00fc\u015flerin belirlenmesi s\u00fcrecidir diyebiliriz.<\/p><\/blockquote>\n Ayr\u0131ca bir konu\u015fmac\u0131n\u0131n d\u00fc\u015f\u00fcncesini veya tutumunu inceleyen, fikir madencili\u011fi olarak da bilinir. \u00d6zellikle machine learning<\/em>‘in h\u0131zla ilerlemesiyle beraber, g\u00fcn\u00fcm\u00fcz teknoloji \u00e7a\u011f\u0131nda sentiment analizleri \u00fczerinde de b\u00fcy\u00fck \u00e7al\u0131\u015fmalar yap\u0131lmaktad\u0131r.. Ge\u00e7mi\u015fe d\u00f6n\u00fcp bir 10 y\u0131l \u00f6ncesini hat\u0131rlarsak, forex firmalar\u0131 i\u00e7erisinde de piyasalar\u0131n sentiment analizlerinin \u00e7\u0131kart\u0131l\u0131p, bu analizler do\u011frultusunda al\u0131m ve sat\u0131m i\u015flemlerinin yap\u0131ld\u0131\u011f\u0131n\u0131 g\u00f6rebiliriz. (Oynayanlar bilir)<\/p>\n G\u00fcn\u00fcm\u00fczde ise firmalar, kendi \u00fcr\u00fcnleri hakk\u0131nda sosyal medya \u00fczerinde ne d\u00fc\u015f\u00fcn\u00fcl\u00fcyor gibi fikirleri \u00f6\u011frenebilmek i\u00e7in, olduk\u00e7a yayg\u0131n olarak sentiment analizi gibi hizmetleri kullanmaktad\u0131rlar.<\/p>\n \u00d6rne\u011fin\u00a0twitter<\/em> veya instagram<\/em> \u00fczerinde yedi\u011fimiz-i\u00e7ti\u011fimiz \u015feyleri, duygular\u0131m\u0131z\u0131 genelde payla\u015fmaktay\u0131z. Bu payla\u015f\u0131m i\u015flemleri basit bir olay gibi g\u00f6z\u00fckebilir, fakat bir \u00e7ok firma hangi \u00fcr\u00fcnlere veya hangi y\u00f6ne a\u011f\u0131rl\u0131k vermeleri gerekti\u011fi bilgilerini bu payla\u015f\u0131mlardan ve analizlerinden yola \u00e7\u0131karak belirleyebilmektedir.<\/p>\n Biz ise bu makale kapsam\u0131nda\u00a0sentiment analizini, kullan\u0131c\u0131lar\u0131n \u00fcr\u00fcnler hakk\u0131nda ne hissettiklerini ke\u015ffedip, bir sonraki kullan\u0131c\u0131ya tercih edece\u011fi \u00fcr\u00fcn hakk\u0131nda daha h\u0131zl\u0131 karar verip, sat\u0131n alabilmesi amac\u0131yla esneklik sa\u011flamak i\u00e7in kullanaca\u011f\u0131z.<\/p>\n Sentiment analizini ger\u00e7ekle\u015ftirebilmenin bir \u00e7ok farkl\u0131 y\u00f6ntemi mevcut. \u00d6rne\u011fin,\u00a0Python<\/em>‘\u0131n VADER<\/em>\u00a0NLTK<\/em><\/a> paketi ile kendi sentiment analyser’\u0131n\u0131z\u0131 olu\u015fturabilirsiniz (\u00dczerinde \u00e7al\u0131\u015ft\u0131\u011f\u0131m bir tool, ilerleyen makale konusu olarak de\u011ferlendirebilirim) veya bizi h\u0131zland\u0131racak bir cloud provider API<\/em>‘\u0131 tercih edebilirsiniz.<\/p>\n Biz bu makale kapsam\u0131nda ise sentiment analizini,\u00a0Azure Text Analytics API<\/em>‘\u0131 kullanarak ger\u00e7ekle\u015ftirece\u011fiz.<\/p>\n Peki nedir bu Azure Text Analytics API <\/em>ve bize neler sunuyor?<\/p>\n Azure Text Analytics API<\/em>, raw text \u00fczerinden\u00a0advanced natural language processing<\/em>(geli\u015fmi\u015f do\u011fal dil i\u015fleme) ger\u00e7ekle\u015ftirebilmemize olanak sa\u011flayan bir cloud-based hizmettir. \u00d6zellikle “time to market<\/em>” s\u00f6z konusu oldu\u011funda, Azure Text Analytics API<\/em> gibi cloud hizmetleri bizlere olduk\u00e7a fazla “h\u0131z<\/strong>” ve “zaman<\/strong>” kazand\u0131rmaktad\u0131r.<\/p>\n Azure Text Analytics API<\/em>, a\u015fa\u011f\u0131daki 4 ana fonksiyonu i\u00e7ermektedir:<\/p>\n NOT<\/strong>: Free tier’\u0131 se\u00e7erek, ayl\u0131k 5000<\/em><\/strong> transaction’a kadar \u00fccretsiz olarak kullanabilmek m\u00fcmk\u00fcnd\u00fcr.<\/p><\/blockquote>\n Text Analytics API<\/em>‘\u0131 kullanabilmek i\u00e7in Azure<\/em>\u00a0marketplace \u00fczerinden “AI + Machine Learning<\/em>” sekmesine girelim ve “Text Analytics<\/em>” i se\u00e7elim.<\/p>\n Ard\u0131ndan a\u015fa\u011f\u0131daki gerekli alanlar\u0131, kendi se\u00e7imlerimize g\u00f6re doldural\u0131m ve create<\/strong> butonuna basal\u0131m.<\/p>\n Art\u0131k API haz\u0131r durumda.<\/p>\n A\u015fa\u011f\u0131daki overview ekran’\u0131 \u00fczerinden Azure Text Analytics API<\/em>‘\u0131n\u0131n, makalenin ilerleyen b\u00f6l\u00fcmlerinde kullanabilmek i\u00e7in “endpoint<\/em>” ve “key<\/em>” bilgilerine eri\u015febilmek m\u00fcmk\u00fcnd\u00fcr.<\/p>\n Bir e-ticaret firmas\u0131nda \u00e7al\u0131\u015ft\u0131\u011f\u0131m\u0131z\u0131 ve kullan\u0131c\u0131lar\u0131n sat\u0131n ald\u0131klar\u0131 \u00fcr\u00fcnler i\u00e7in yorum yapabildiklerini varsayal\u0131m. San\u0131r\u0131m bir \u00fcr\u00fcn\u00fc sat\u0131n almadan \u00f6nce o \u00fcr\u00fcn i\u00e7in yap\u0131lm\u0131\u015f olan yorumlar\u0131 okuyabilmek, hem son kullan\u0131c\u0131 a\u00e7\u0131s\u0131ndan hem de ilgili firma a\u00e7\u0131s\u0131ndan \u00f6nemli bir fonksiyonalitedir.<\/p>\n Peki, son kullan\u0131c\u0131lar\u0131n t\u00fcm yorumlar\u0131 okuyup bir \u00fcr\u00fcn se\u00e7imi yapmalar\u0131n\u0131n yerine, biz onlar i\u00e7in t\u00fcm \u00fcr\u00fcn yorumlar\u0131 \u00fczerinde bir sentiment analizi ger\u00e7ekle\u015ftirerek, her bir \u00fcr\u00fcn i\u00e7in ortalama bir son kullan\u0131c\u0131 puan\u0131 g\u00f6sterebilsek, harika olmaz m\u0131? Hem son kullan\u0131c\u0131lar t\u00fcm yorumlar\u0131 okuyarak fazla vakit kaybetmemi\u015f olurlar hem de biz son kullan\u0131c\u0131n\u0131n ziyaretini h\u0131zl\u0131ca sat\u0131\u015fa \u00e7evirebilme imkan\u0131 elde etmi\u015f olabiliriz.<\/p>\n Haydi ozaman i\u015f ba\u015f\u0131na.<\/p>\n \u0130lk olarak a\u015fa\u011f\u0131daki gibi “SentimentAnalysisWithNETCoreExample<\/em>” isminde bir .NET Core<\/em>\u00a0“webapi<\/em>” projesi olu\u015ftural\u0131m.<\/p>\n Ard\u0131ndan projeye a\u015fa\u011f\u0131daki komut sat\u0131r\u0131 ile “Microsoft.EntityFrameworkCore<\/em>” paketini dahil edelim.<\/p>\n Art\u0131k domain modellerimizi tan\u0131mlayabiliriz.<\/p>\n “Models<\/em>” isminde bir klas\u00f6r olu\u015fturarak, i\u00e7erisinde “Domain<\/em>” isminde yeni bir klas\u00f6r daha olu\u015ftural\u0131m.<\/p>\n \u015eimdi “Domain<\/em>” klas\u00f6r\u00fc i\u00e7erisinde ise, a\u015fa\u011f\u0131daki gibi bir “Product<\/em>” class’\u0131 ekleyelim.<\/p>\n Bizim i\u00e7in buradaki \u00f6nemli nokta, “CustomerRating<\/em>” property’si. Bu property’nin value’sunu, product comment’lerinin sentiment analizi sonucunda \u00e7\u0131kan score’lar\u0131n ortalamas\u0131 ile dolduraca\u011f\u0131z.<\/p>\n \u015eimdi ise “Domain<\/em>” klas\u00f6r\u00fc i\u00e7erisine, “Comment<\/em>” isminde bir class daha ekleyelim.<\/p>\n Bu class i\u00e7erisindeki “SentimentScore<\/em>” property’sinin value’sunda ise, her bir comment’in kendi sentiment score’unu tutaca\u011f\u0131z.<\/p>\n Art\u0131k data context’i ve sample dataset’i olu\u015fturabiliriz. “Data<\/em>” isminde root dizinde bir klas\u00f6r olu\u015ftural\u0131m ve i\u00e7erisinde “ProductDBContext<\/em>” isminde a\u015fa\u011f\u0131daki gibi bir class ekleyelim.<\/p>\n Data context’i standart olarak “DbContext<\/em>” class’\u0131ndan inherit alarak olu\u015fturduk. \u0130\u00e7erisinde “Products<\/em>” ve “Comments<\/em>” dbset’lerini tan\u0131mlad\u0131k. \u00d6rnek dataset’e sahip olabilmek i\u00e7in, “OnModelCreating<\/em>” method’u i\u00e7erisinde bir ka\u00e7 product ve onlara comment’ler ekledik.<\/p>\n \u015eimdi ise, business service’leri kodlamaya ba\u015flamadan \u00f6nce request & response modellerimizi tan\u0131mlayal\u0131m.<\/p>\n Bunun i\u00e7in “Models<\/em>” klas\u00f6r\u00fc alt\u0131nda, “Requests<\/em>” ve “Responses<\/em>” klas\u00f6r\u00fclerini olu\u015ftural\u0131m. Ard\u0131ndan “Requests<\/em>” klas\u00f6r\u00fcn\u00fcn i\u00e7erisinde a\u015fa\u011f\u0131daki gibi “GetSentimentAnalysisRequest<\/em>” ve “GetSentimentAnalysisRequestItem<\/em>” class’lar\u0131n\u0131 tan\u0131mlayal\u0131m.<\/p>\n “GetSentimentAnalysisRequest<\/em>” ve “GetSentimentAnalysisRequestItem<\/em>” class’lar\u0131n\u0131, Azure Text Analytics API<\/em>‘\u0131n sentiment endpoint’ini call edebilmek i\u00e7in kullanaca\u011f\u0131z.<\/p>\n Sentiment endpoint’i a\u015fa\u011f\u0131daki gibi bir request’i bizden beklemektedir:<\/p>\n NOT<\/strong>: Azure Text Analytics API<\/em>‘\u0131n sentiment endpoint’ini kullanabilmek i\u00e7in haz\u0131r bir NuGet<\/em><\/a> package’\u0131 mevcut. Fakat preview durumdad\u0131r ve hen\u00fcz\u00a0.NET Standard<\/em> 1.4’\u00fc desteklemektedir. Bu sebeple\u00a0.NET Core 2.1<\/em> uygulamas\u0131nda kullanabilmek i\u00e7in, kendi request modellerimizi ve service’imizi implemente edece\u011fiz.<\/p><\/blockquote>\n Sentiment endpoint’inin response modellerini ise, olu\u015fturmu\u015f oldu\u011fumuz “Responses<\/em>” klas\u00f6r\u00fc alt\u0131nda tan\u0131mlayaca\u011f\u0131z.<\/p>\n “GetSentimentAnalysisResponse<\/em>” ve “GetSentimentAnalysisResponseItem<\/em>” ismindeki iki adet class’\u0131, “Responses<\/em>” klas\u00f6r\u00fc alt\u0131nda a\u015fa\u011f\u0131daki gibi olu\u015ftural\u0131m.<\/p>\n Art\u0131k sentiment sonu\u00e7lar\u0131n\u0131 alabilece\u011fimiz response modellerini de tan\u0131mlad\u0131\u011f\u0131m\u0131za g\u00f6re, service’i implemente etmeye ba\u015flayabiliriz.<\/p>\n Bunun i\u00e7in root dizinde “Services<\/em>” isminde yeni bir klas\u00f6r olu\u015ftural\u0131m ve i\u00e7erisinde “ITextAnalyticsService<\/em>” isminde bir interface tan\u0131mlayal\u0131m.<\/p>\n Ard\u0131ndan “Services<\/em>” klas\u00f6r\u00fc alt\u0131nda “Implementations<\/em>” isminde bir yeni klas\u00f6r daha olu\u015ftural\u0131m ve i\u00e7erisine “TextAnalyticsService<\/em>” isminde bir class yarat\u0131p, “ITextAnalyticsService<\/em>” interface’ini a\u015fa\u011f\u0131daki gibi implemente edelim.<\/p>\n “TextAnalyticsService<\/em>” i\u00e7erisinde yapt\u0131\u011f\u0131m\u0131z i\u015flem gayet basit.<\/p>\n HttpClientFactory<\/em>‘i “TextAnalyticsAPI<\/em>” key’i \u00fczerinden named-client y\u00f6ntemi ile olu\u015fturduk. Ard\u0131ndan “IConfiguration<\/em>” service’i \u00fczerinden ilgili Azure Text Analytics API<\/em>‘\u0131n sentiment resource URI<\/em>‘\u0131n\u0131 alarak, POST<\/em> i\u015flemini ger\u00e7ekle\u015ftirdik. E\u011fer i\u015flem ba\u015far\u0131l\u0131ysa tamamlan\u0131yorsa, response’u daha \u00f6nce olu\u015fturmu\u015f oldu\u011fumuz “GetSentimentAnalysisResponse<\/em>” model’i ile map’liyoruz.<\/p>\n \u015eimdi ise product comment’leri ile ilgili i\u015flemleri implemente edece\u011fimiz service’in interface’ini, a\u015fa\u011f\u0131daki gibi “Services<\/em>” klas\u00f6r\u00fc alt\u0131nda tan\u0131mlayal\u0131m.<\/p>\n “IProductCommentService<\/em>”<\/p>\n Ard\u0131ndan “Services<\/em>” klas\u00f6r\u00fc alt\u0131ndaki “Implementations<\/em>” klas\u00f6r\u00fc i\u00e7erisinde, “ProductCommentService<\/em>” isminde bir class daha olu\u015ftural\u0131m ve “IProductCommentService<\/em>” interface’ini a\u015fa\u011f\u0131daki gibi implemente edelim.<\/p>\n “GetCommentsAsync<\/em>” method’u ile ilgili product’\u0131n comment’lerini, database’den al\u0131yoruz. “CalculateCommentsSentimentScoreAsync<\/em>” method’u i\u00e7erisinde ise olu\u015fturmu\u015f oldu\u011fumuz Azure Text Analytics API<\/em>‘\u0131n service’ini kullanarak, t\u00fcm comment’lerin sentiment score’lar\u0131n\u0131 hesapl\u0131yoruz. E\u011fer API<\/em> call’\u0131 s\u0131ras\u0131nda herhangi bir problem olu\u015fmazsa, comment’lerin sentiment score’lar\u0131n\u0131 map’liyoruz.<\/p>\n \u015eimdi ise product’lar ile ilgili i\u015flemleri ger\u00e7ekle\u015ftirebilece\u011fimiz bir service’e daha ihtiyac\u0131m\u0131z var.<\/p>\n \u00d6ncelikle d\u0131\u015far\u0131ya domain model’ini expose etmemek ad\u0131na product response modelini, daha \u00f6nce olu\u015fturmu\u015f oldu\u011fumuz “Models\/Responses<\/em>” klas\u00f6r\u00fc i\u00e7erisinde tan\u0131mlamam\u0131z gerekmektedir.<\/p>\n “GetProductResponse<\/em>” ve “GetProductCommentResponse<\/em>” class’lar\u0131n\u0131 a\u015fa\u011f\u0131daki gibi “Responses<\/em>” klas\u00f6r\u00fc i\u00e7erisinde tan\u0131mlayal\u0131m.<\/p>\n Model’leri tan\u0131mlaman\u0131n ard\u0131ndan, “Services<\/em>” klas\u00f6r\u00fc alt\u0131nda “IProductService<\/em>” isminde a\u015fa\u011f\u0131daki gibi yeni bir interface olu\u015ftural\u0131m.<\/p>\n Ard\u0131ndan “Services<\/em>” klas\u00f6r\u00fc alt\u0131ndaki “Implementations<\/em>” klas\u00f6r\u00fc i\u00e7erisinde, “ProductService<\/em>” isminde bir class daha olu\u015fturarak, a\u015fa\u011f\u0131daki gibi implemente edelim.<\/p>\n![\"\"](\"\/wp-content\/uploads\/2018\/07\/sentiment-analysis.jpg\")
<\/a><\/p>\n1) Sentiment<\/em> Analizi Nedir?<\/h2>\n
<\/a><\/p>\n2)\u00a0Azure Text Analytics API<\/em><\/h2>\n
<\/a><\/p>\n\n
2.1)\u00a0Text Analytics Resource’unu Olu\u015fturmak<\/em><\/h3>\n
<\/a><\/p>\n<\/a><\/p>\n<\/a><\/p>\n3) .NET Core ile\u00a0Azure Text Analytics API Kullan\u0131m\u0131<\/em><\/h2>\n
<\/a><\/p>\ndotnet new webapi -n SentimentAnalysisWithNETCoreExample<\/pre>\n
<\/a><\/p>\ndotnet add package Microsoft.EntityFrameworkCore<\/pre>\n
-> SentimentAnalysisWithNETCoreExample\r\n--> Models\r\n---> Domain<\/pre>\n
using System.Collections.Generic;\r\n\r\nnamespace SentimentAnalysisWithNETCoreExample.Models.Domain\r\n{\r\n public class Product\r\n {\r\n public Product()\r\n {\r\n this.Comments = new List<Comment>();\r\n }\r\n\r\n public int ID { get; set; }\r\n public string Name { get; set; }\r\n public string Category { get; set; }\r\n public double Price { get; set; }\r\n public double CustomerRating { get; set; }\r\n\r\n public ICollection<Comment> Comments { get; set; }\r\n }\r\n}<\/pre>\nnamespace SentimentAnalysisWithNETCoreExample.Models.Domain\r\n{\r\n public class Comment\r\n {\r\n public int ID { get; set; }\r\n public int ProductID { get; set; }\r\n public string Content { get; set; }\r\n public double? SentimentScore { get; set; }\r\n }\r\n}<\/pre>\nusing Microsoft.EntityFrameworkCore;\r\nusing SentimentAnalysisWithNETCoreExample.Models.Domain;\r\n\r\nnamespace SentimentAnalysisWithNETCoreExample.Data\r\n{\r\n public class ProductDBContext : DbContext\r\n {\r\n public ProductDBContext(DbContextOptions<ProductDBContext> options)\r\n : base(options)\r\n {\r\n\r\n }\r\n\r\n public DbSet<Product> Products { get; set; }\r\n public DbSet<Comment> Comments { get; set; }\r\n\r\n protected override void OnModelCreating(ModelBuilder modelBuilder)\r\n {\r\n #region Products\r\n var product_1 = new Product()\r\n {\r\n ID = 1,\r\n Name = \"Samsung Note 9\",\r\n Category = \"Cep Telefonu\",\r\n Price = 600\r\n };\r\n\r\n var product_2 = new Product()\r\n {\r\n ID = 2,\r\n Name = \"Samsung Galaxy 9\",\r\n Category = \"Cep Telefonu\",\r\n Price = 550\r\n };\r\n\r\n modelBuilder.Entity<Product>().HasData(product_1, product_2);\r\n #endregion\r\n\r\n #region Comments\r\n var comment_1 = new Comment()\r\n {\r\n ID = 1,\r\n ProductID = 1,\r\n Content = \"\u00c7ok harika bir telefon. \u00c7ok memnunum. Herkese tavsiye ederim.\"\r\n };\r\n\r\n var comment_2 = new Comment()\r\n {\r\n ID = 2,\r\n ProductID = 1,\r\n Content = \"G\u00fczel telefon, gayet kullan\u0131\u015fl\u0131.\"\r\n };\r\n\r\n var comment_3 = new Comment()\r\n {\r\n ID = 3,\r\n ProductID = 1,\r\n Content = \"H\u0131zl\u0131 telefon, be\u011fendim.\"\r\n };\r\n\r\n var comment_4 = new Comment()\r\n {\r\n ID = 4,\r\n ProductID = 1,\r\n Content = \"Bataryas\u0131 \u00e7ok iyi dayan\u0131yor tavsiye ederim.\"\r\n };\r\n\r\n var comment_5 = new Comment()\r\n {\r\n ID = 5,\r\n ProductID = 2,\r\n Content = \"Fena de\u011fil. Biraz \u0131s\u0131n\u0131yor. Daha iyi olabilirdi.\"\r\n };\r\n\r\n var comment_6 = new Comment()\r\n {\r\n ID = 6,\r\n ProductID = 2,\r\n Content = \"Kasas\u0131n\u0131 be\u011fenmedim. Elden \u00e7ok kay\u0131yor.\"\r\n };\r\n\r\n var comment_7 = new Comment()\r\n {\r\n ID = 7,\r\n ProductID = 2,\r\n Content = \"Memnunum, kameras\u0131 \u00e7ok iyi.\"\r\n };\r\n\r\n var comment_8 = new Comment()\r\n {\r\n ID = 8,\r\n ProductID = 2,\r\n Content = \"Kameras\u0131n\u0131n gece \u00e7ekimini be\u011fenmedim.\"\r\n };\r\n\r\n modelBuilder.Entity<Comment>().HasData(comment_1, comment_2, comment_3, comment_4, comment_5, comment_6, comment_7, comment_8);\r\n #endregion\r\n\r\n base.OnModelCreating(modelBuilder);\r\n }\r\n }\r\n}<\/pre>\nusing System.Collections.Generic;\r\n\r\nnamespace SentimentAnalysisWithNETCoreExample.Models.Requests\r\n{\r\n public class GetSentimentAnalysisRequest\r\n {\r\n public GetSentimentAnalysisRequest()\r\n {\r\n Documents = new List<GetSentimentAnalysisRequestItem>();\r\n }\r\n\r\n public IList<GetSentimentAnalysisRequestItem> Documents { get; set; }\r\n }\r\n}<\/pre>\nnamespace SentimentAnalysisWithNETCoreExample.Models.Requests\r\n{\r\n public class GetSentimentAnalysisRequestItem\r\n {\r\n public string Language { get; set; }\r\n public string Id { get; set; }\r\n public string Text { get; set; }\r\n }\r\n}<\/pre>\n{\r\n \"documents\": [\r\n {\r\n \"language\": \"comment lang\",\r\n \"id\": \"comment id\",\r\n \"text\": \"comment\"\r\n },\r\n {\r\n \"language\": \"comment lang\",\r\n \"id\": \"comment id\",\r\n \"text\": \"comment\"\r\n }\r\n ]\r\n}<\/pre>\nusing System.Collections.Generic;\r\n\r\nnamespace SentimentAnalysisWithNETCoreExample.Models.Responses\r\n{\r\n public class GetSentimentAnalysisResponse\r\n {\r\n public IList<GetSentimentAnalysisResponseItem> Documents { get; set; }\r\n }\r\n}<\/pre>\nnamespace SentimentAnalysisWithNETCoreExample.Models.Responses\r\n{\r\n public class GetSentimentAnalysisResponseItem\r\n {\r\n public string Id { get; set; }\r\n public double? Score { get; set; }\r\n }\r\n}<\/pre>\nusing System.Threading.Tasks;\r\nusing SentimentAnalysisWithNETCoreExample.Models.Requests;\r\nusing SentimentAnalysisWithNETCoreExample.Models.Responses;\r\n\r\nnamespace SentimentAnalysisWithNETCoreExample.Services\r\n{\r\n public interface ITextAnalyticsService\r\n {\r\n Task<GetSentimentAnalysisResponse> GetSentimentAsync(GetSentimentAnalysisRequest request);\r\n }\r\n}<\/pre>\nusing System.Net;\r\nusing System.Net.Http;\r\nusing System.Threading.Tasks;\r\nusing Microsoft.Extensions.Configuration;\r\nusing Newtonsoft.Json;\r\nusing SentimentAnalysisWithNETCoreExample.Models.Requests;\r\nusing SentimentAnalysisWithNETCoreExample.Models.Responses;\r\n\r\nnamespace SentimentAnalysisWithNETCoreExample.Services.Implementations\r\n{\r\n public class TextAnalyticsService : ITextAnalyticsService\r\n {\r\n private readonly IHttpClientFactory _httpClientFactory;\r\n private readonly IConfiguration _configuration;\r\n\r\n public TextAnalyticsService(IHttpClientFactory httpClientFactory, IConfiguration configuration)\r\n {\r\n _httpClientFactory = httpClientFactory;\r\n _configuration = configuration;\r\n }\r\n\r\n public async Task<GetSentimentAnalysisResponse> GetSentimentAsync(GetSentimentAnalysisRequest request)\r\n {\r\n HttpClient client = _httpClientFactory.CreateClient(\"TextAnalyticsAPI\");\r\n\r\n var sentimentResponse = await client.PostAsJsonAsync(\r\n requestUri: _configuration.GetValue<string>(\"TextAnalyticsAPISentimentResourceURI\"), \r\n value: request);\r\n\r\n GetSentimentAnalysisResponse getSentimentAnalysisResponse = null;\r\n\r\n if(sentimentResponse.StatusCode == HttpStatusCode.OK)\r\n {\r\n getSentimentAnalysisResponse = await sentimentResponse.Content.ReadAsAsync<GetSentimentAnalysisResponse>();\r\n }\r\n\r\n return getSentimentAnalysisResponse;\r\n }\r\n }\r\n}<\/pre>\nusing System.Collections.Generic;\r\nusing System.Threading.Tasks;\r\nusing SentimentAnalysisWithNETCoreExample.Models.Domain;\r\n\r\nnamespace SentimentAnalysisWithNETCoreExample.Services\r\n{\r\n public interface IProductCommentService\r\n {\r\n Task<List<Comment>> GetComments(int productId);\r\n Task<List<Comment>> CalculateCommentsSentimentScoreAsync(List<Comment> comments);\r\n }\r\n}<\/pre>\nusing System;\r\nusing System.Collections.Generic;\r\nusing System.Linq;\r\nusing System.Threading.Tasks;\r\nusing Microsoft.EntityFrameworkCore;\r\nusing SentimentAnalysisWithNETCoreExample.Data;\r\nusing SentimentAnalysisWithNETCoreExample.Models.Domain;\r\nusing SentimentAnalysisWithNETCoreExample.Models.Requests;\r\nusing SentimentAnalysisWithNETCoreExample.Models.Responses;\r\n\r\nnamespace SentimentAnalysisWithNETCoreExample.Services.Implementations\r\n{\r\n public class ProductCommentService : IProductCommentService\r\n {\r\n private const string LANG = \"tr\";\r\n private readonly ProductDBContext _productDBContext;\r\n private readonly ITextAnalyticsService _textAnalyticsService;\r\n\r\n public ProductCommentService(ProductDBContext productDBContext, ITextAnalyticsService textAnalyticsService)\r\n {\r\n _productDBContext = productDBContext;\r\n _textAnalyticsService = textAnalyticsService;\r\n }\r\n\r\n public async Task<List<Comment>> GetComments(int productId)\r\n {\r\n List<Comment> comments = await _productDBContext.Comments.Where(c => c.ProductID == productId)\r\n .ToListAsync();\r\n\r\n return comments;\r\n }\r\n\r\n public async Task<List<Comment>> CalculateCommentsSentimentScoreAsync(List<Comment> comments)\r\n {\r\n var getSentimentAnalysisRequest = new GetSentimentAnalysisRequest();\r\n\r\n comments.ForEach(comment =>\r\n {\r\n var multiLanguageInput = new GetSentimentAnalysisRequestItem()\r\n {\r\n Language = LANG,\r\n Id = comment.ID.ToString(),\r\n Text = comment.Content\r\n };\r\n\r\n getSentimentAnalysisRequest.Documents.Add(multiLanguageInput);\r\n });\r\n\r\n GetSentimentAnalysisResponse getSentimentAnalysisResponse = await _textAnalyticsService.GetSentimentAsync(getSentimentAnalysisRequest);\r\n\r\n if (getSentimentAnalysisResponse != null && getSentimentAnalysisResponse.Documents.Count > 0)\r\n {\r\n \/\/ Add sentiment analysis result to the comments\r\n foreach (GetSentimentAnalysisResponseItem getSentimentAnalysisResponseItem in getSentimentAnalysisResponse.Documents)\r\n {\r\n Comment comment = comments.FirstOrDefault(c => c.ID == Convert.ToInt32(getSentimentAnalysisResponseItem.Id));\r\n\r\n comment.SentimentScore = getSentimentAnalysisResponseItem.Score;\r\n }\r\n }\r\n\r\n return comments;\r\n }\r\n }\r\n}<\/pre>\nusing System.Collections.Generic;\r\n\r\nnamespace SentimentAnalysisWithNETCoreExample.Models.Responses\r\n{\r\n public class GetProductResponse\r\n {\r\n public int ID { get; set; }\r\n public string Name { get; set; }\r\n public string Category { get; set; }\r\n public double Price { get; set; }\r\n public double CustomerRating { get; set; }\r\n public List<GetProductCommentResponse> Comments { get; set; }\r\n }\r\n}<\/pre>\nnamespace SentimentAnalysisWithNETCoreExample.Models.Responses\r\n{\r\n public class GetProductCommentResponse\r\n {\r\n public string Content { get; set; }\r\n }\r\n}<\/pre>\nusing System.Collections.Generic;\r\nusing System.Threading.Tasks;\r\nusing SentimentAnalysisWithNETCoreExample.Models.Domain;\r\nusing SentimentAnalysisWithNETCoreExample.Models.Responses;\r\n\r\nnamespace SentimentAnalysisWithNETCoreExample.Services\r\n{\r\n public interface IProductService\r\n {\r\n Task<GetProductResponse> GetProductAsync(int id);\r\n }\r\n}<\/pre>\nusing System.Collections.Generic;\r\nusing System.Linq;\r\nusing System.Threading.Tasks;\r\nusing Microsoft.EntityFrameworkCore;\r\nusing SentimentAnalysisWithNETCoreExample.Data;\r\nusing SentimentAnalysisWithNETCoreExample.Models.Domain;\r\nusing SentimentAnalysisWithNETCoreExample.Models.Responses;\r\n\r\nnamespace SentimentAnalysisWithNETCoreExample.Services.Implementations\r\n{\r\n public class ProductService : IProductService\r\n {\r\n private readonly ProductDBContext _productDBContext;\r\n private readonly IProductCommentService _productCommentService;\r\n\r\n public ProductService(ProductDBContext productDBContext, IProductCommentService productCommentService)\r\n {\r\n _productDBContext = productDBContext;\r\n _productCommentService = productCommentService;\r\n }\r\n\r\n public async Task<GetProductResponse> GetProductAsync(int id)\r\n {\r\n Product product = await _productDBContext.Products.Where(p => p.ID == id)\r\n .FirstOrDefaultAsync();\r\n\r\n GetProductResponse productResponse = null;\r\n\r\n if (product != null)\r\n {\r\n productResponse = new GetProductResponse();\r\n \r\n productResponse.ID = product.ID;\r\n productResponse.Name = product.Name;\r\n productResponse.Category = product.Category;\r\n productResponse.Price = product.Price;\r\n\r\n List<Comment> comments = await _productCommentService.GetComments(id);\r\n\r\n productResponse.Comments = comments.Select(c => new GetProductCommentResponse()\r\n {\r\n Content = c.Content\r\n }).ToList();\r\n\r\n if (comments.Count > 0)\r\n {\r\n List<Comment> commentsWithSentimentAnalysis = await _productCommentService.CalculateCommentsSentimentScoreAsync(comments);\r\n\r\n productResponse.CustomerRating = await CalculateProductCustomerRatingScore(commentsWithSentimentAnalysis);\r\n }\r\n }\r\n\r\n return productResponse;\r\n }\r\n\r\n private async Task<double> CalculateProductCustomerRatingScore(List<Comment> comments)\r\n {\r\n double sentimentScores = 0;\r\n double customerRating = 0;\r\n\r\n comments.ForEach(_comment =>\r\n {\r\n sentimentScores += _comment.SentimentScore.Value;\r\n });\r\n\r\n customerRating = (sentimentScores \/ comments.Count());\r\n\r\n return await Task.FromResult(customerRating);\r\n }\r\n }\r\n}<\/pre>\n
![](\"\/wp-content\/uploads\/2018\/12\/sentiment-analysis-olumlu.png\"){kind=link}