Sentiment analysis and aspect based on product reviews by using unsupervised learning Harvard Case Solution & Analysis

Sentiment analysis and aspect based on product reviews by using unsupervised learning Case Study Solution

‘I enjoyed the screen resolution, it is amazing for such a cheap laptop.’

ABSA – Aspect based sentiment analysis provides assistance to businesses to transform intocustomer-centric with placement of their potential customers at the highest level of their priority list. It is basically all about listening their consumers, a better understanding of their voices, their feedback analysis as well as learning about experiences of customers for different and distinctive products and services.

While, computation approaches utilization in the mining customer opinion tend to provide proof of being promising, as it indicates the things that are required for significant performance improvement of such models. One of the plausible approach in terms of performance improvement is expected for future use with advanced techniques of word presentation particularly contextual embedding of the words.

According to a study conducted by Samuel Brody et al. on unsupervised ABSA for product review confirmed the value of a completely unsupervised approach on the basis of the tasks of detection of aspect and sentiment analysis. The inference of such aspects were primarily from the data which are considered more representative in nature in comparison to ones that are derived manually. As an illustration, in the domain of restaurant, the aspect list which involves construction by manual means omitted or over-generalized a number of important aspects. However, it had a strong presence in the data. The aspect of service which deals with mistaken orders, reservations and waiting time was thought to be an important aspect of emergency on its own. But, it was grouped in the manual annotation under staff.

Deliverance of different sentiments through use of adjectives depends on the aspect that requires discussion in future. For instance, the ranking of adjective ‘warm’ was actually considered quite positive in the aspect of staff but with a slight negative influence in the aspect of General Food. An approach i.e. knowledge-richis expected with ignorance of such adjectives, therebymissing some important review elements.

At last, due to the belongings of online reviews to genre that are informal with specialized jargon andinventive spelling, it might not be sufficient for both sentiment and aspect in order to completely rely on lexicons. For instance, the reviews of restaurants tend to involve errors of spelling such as tastey, exelent, omelete, sandwhich, creme-brule, anti-pasta, decour/decore and desert as well as a number of no less than six different common restaurant misspellings. The availability of some special terminologies such as Pho, Dosa, Edamame and Korma does not seem to appear in general dictionaries, and creative adjective use like New-Yorky and orgasmic.

Such research studies have significantly opened many avenues for the purpose of future improvements in research studies. However, the primary focus had known primarily on adjectives as indicators of sentiment. Although, there are a number of studies with use of other parts of speech that can provide assistance in such tasks.


Exhibit A – Sentiments of Sentiment Analysis

Exhibit B –Basic Classification

Exhibit C –Tasks of ABSA


This is just a sample partical work. Please place the order on the website to get your own originally done case solution.

How We Work?
Just email us your case materials and instructions to [email protected] and confirm your order by making the payment here

Other Similar Case Solutions like

Sentiment analysis and aspect based on product reviews by using unsupervised learning

Share This