A Confirmatory Topic Matching for Online Reviews

A new opinion extraction method is proposed to summarize unstructured user-generated contents (i.e., customer online reviews) into fixed topic domains. To differentiate the current approach from other opinion extraction approaches often exposed to sparsity problem and lack of sentiment scores, a confirmatory topic matching (CTM) framework is introduced along with its practical algorithm called DiSBUSS. The proposed CTM is applied to customer reviews of restaurants in Hawaii obtained from TripAdvisor and the empirical findings confirm the effectiveness or validity of the method.

Customer Perception on Smart Home Appliances Based on Amazon Reviews

It is difficult to capture residents’ perception or opinion on energy conservation practice. A simple way is to conduct a survey, but it often requires a lot of costs. We discuss an opinion mining algorithm (Aspect-based opinion mining based on natural language processing) that extract residents’ opinion through the pre-selected energy product reviews.

Latent Dirichlet Allocation Models Considering Emojis

The rapid growth in internet and telecommunication technology triggered the development of Social Networking Service (SNS) platforms. This electronic word-of-mouth heavily uses text data as the medium of communication. An ideal method for analyzing SNS text data for information retrieval tasks is Latent Dirichlet Allocation (LDA). LDA exploits statistical inference to discover structure given the words and documents by calculating the relative importance of topics in documents and words in topics. Conventionally, emoji - a pictogram that expresses the author’s feeling and emotion - characters have been considered as a noise and were deleted prior to applying LDA techniques and other topic modeling methods. The paper, on the other hand, insist that one should focus on the richness of information that emoji characters can provide.

Download pdf

A Translation Approach for Unstructured Online Reviews

Build structure on online reviews by changing topic key words in each review to the pre-assumed words by linking online dictionary and then decompose each review into several components, sentences or phrases, to include only one topic in a component. After then, each component is scored using a machine learning method. The scored data are used to analysis customers' satisfaction in terms of assumed topic dimensions.

Album Recommendation System Based on Random Forest Method and Mixed Effect Model

An improved music recommendation algorithm based on the content based filtering method by introducing a new variable produced by random mixed effect model and lyric score. The new variable reflects the effect of the artist that cannot be explained by the specific variables, including genre, member composition, and internal information of the album. The lyrics score measures positivity of the words in lyrics. Then, the random forest method will be applied to estimate the review rate of the albums by using the training data based on the billboard chart

Estimate the trend of social atmosphere in the US by examining popular songs played over

This study examines the estimation the trend of social atmosphere in the US using of data mining techniques, especially text mining. Data mining techniques has played an important role in many fields such as marketing, retail, finance, manufacture because handling and storage of huge data has become possible with the development of information and technique. From the songs' lyrics, there is positive or negative meaning difference among years and singers. Key words: Lyrics, Singer, Song, Billboard, Random Number Simulation.

Vignette

Emoji

Emoji is an R package. Emoji package contains infromation of the official Unicode Emoji v5.0, and functions and codes that could be used in text analytics and sentiment analysis considering Emoji.

Relp

Relp is an application that reconstructs the popular online platform `yelp.com` onto a R Shiny-based application. Relp project was an excellent excercise for web scraping, cleaning, application development.