аннотация6 (Аннотации)
Описание файла
Файл "аннотация6" внутри архива находится в следующих папках: Аннотации, 1. PDF-файл из архива "Аннотации", который расположен в категории "". Всё это находится в предмете "английский язык" из 10 семестр (2 семестр магистратуры), которые можно найти в файловом архиве МГУ им. Ломоносова. Не смотря на прямую связь этого архива с МГУ им. Ломоносова, его также можно найти и в других разделах. .
Просмотр PDF-файла онлайн
Текст из PDF
Kotelnikov E. V., Bushmeleva N. A., Razova E. V., Peskisheva T. A., PletnevaM. V.:Manually Created Sentiment Lexicons: Research and Development.The sentiment lexicons are an important part of many sentiment analysissystems. There are many automatic ways to build such lexicons, but often they aretoo large and contain errors. The manual, dictionary-based and corpus-basedapproaches are typically applied.
They use human labeling, universal dictionariesand thesauri, analysis of text corpora respectively.The paper presents the algorithm of sentiment lexicons creation for a givendomain based on hybrid, manual and corpus-based, approach. This algorithm is usedfor the development of the sentiment lexicons by means of four human annotatorseach for five domains – user reviews of restaurants, cars, movies, books and digitalcameras. Created sentiment lexicons are analyzed for inter-annotator agreement,parts of speech distribution and correlation with automatic lexicons.
At the first stagethe authors retrieve the morphological information of training corpus. Resultingwords are weighted using the Relevance Frequency weighting scheme both forpositive and negative categories. First P words from each list are chosen so that2P = N, where N – a number of words for manual annotation.
Then annotators markwords with one of four labels: positive, negative, neutral and unclear. The finallexicon consists of the words which assessment is accepted by the majority ofannotators.The performance of the sentiment analysis based on the created sentimentlexicons is researched and compared with the performance of the existing sentimentlexicons. The experiments with text corpora on various domains based on SVMshow high quality and compactness of the human-built lexicons..