Skip to main content

Predicting the Future Based on Voice of the Customer (VoC)

Sentiment Analysis Service

Hitachi's Sentiment Analysis Service provides
a highly accurate visual representation of customers' opinions and sentiments
about a company or a product, based on analyses of text data.
The service can analyze the data from a variety of media, such as social media,
customer reviews, mass media including newspapers and television,
and business data from questionnaires and call centers.

And then supporting future planning activities like sales forecasting.

Overview of Sentiment Analysis Service

[Text data from various sources, such as Twitter, surveys, and call centers]→[Analyzing used Sentiment Analysis Service] Sentiment Analysis: Analyze and visualize sentiment with high accuracy based on the meaning of the text. Moral Analysis: Quantitatively analyze and visualize moral values. Insight Analysis: Discover, analyze, and visualize unexpected, previously overlooked information that was buried in huge amounts of data.→[Make new discoveries by combining the results of analyzing Sentiment × Morals × Insight] (Unexpected product needs, Understand causes of social media flaming, uncover risks, and more)→[The Sentiment Analysis Service helps you expand your business and improve your corporate value]

Sentiment Analysis

This function classifies the results into three major emotional categories (positive, negative, and neutral). The text data can then be further broken down into more detailed sentiments (joy, happiness, love, etc.). This function can also classify text into different categories (related to vehicles, food, sports, etc.).

Supported languages: English, Japanese, Thai, Chinese, etc.

Moral Analysis

This function uses the academically backed Moral Foundations Dictionary* to quantitatively analyze and visualize text data and to categorize keywords according to the following five moral foundations: care, fairness, ingroup, authority, and purity.

 *
Moral Foundations Dictionary: Created based on the moral foundations theory put forth by social psychologist Jonathan Haidt, who posits that humans are born with five universal and cross-cultural foundations related to moral right and wrong.
This service uses the following Moral Foundations Dictionaries: MFD and J-MFD.

Supported languages: English, Japanese

Insight Analysis

This function uses Hitachi's original algorithm to identify unexpected, previously overlooked information (such the minority response and as small changes in response) that was buried in huge amounts of data.

Supported languages: English, Japanese

Features of the Service

Facilitates discovery through multifaceted analyses

By combining the Sentiment Analysis, Moral Analysis, and Insight Analysis functions, you can visualize, quantitatively and from various angles, both the majority and minority response from among huge amounts of data (such as data from social media). This allows you to understand values and hidden needs, so you can improve your existing businesses or get ideas for new businesses.

Insightful Viewer

To use sentiment data for your business, you must be able to quickly and easily find the data you want. To achieve this, this service's dashboard has three features to help you notice important points. First, the dashboard displays analysis data in various graphs (pie chart, timeline, word ranking, etc.). Second, the dashboard can filter analysis data not only by keywords and dates, but also by emotions and topics categorized by AI. Finally, the dashboard can also filter analysis data by tags set by users.

Easy Maintenance

Through the machine learning of highly relevant words and technical terms contained in collected data, the AI automatically updates the refining conditions used for the filtering dictionary. This makes it possible to maintain and improve the accuracy of search refinement without additional maintenance work.

Hitachi's Sentiment Analysis Service is available in a default service package or customized systems to meet your needs.
Data from Twitter and Pantip (Thai only) is supported for analysis in the default plan.

01

Product planning to comprehensively meet market needs

category :
  • Planning
  • Development

Perform sentiment analysis of data that includes both business data, such as product functions and performance, and information from social media. Based on customer complaints about the product, this analysis clarifies areas in need of improvement, which allows you to identify customer needs. This, in turn, enables you to perform product planning in a way that comprehensively meets market needs, thereby contributing to your efforts to increase sales and reduce lost opportunities.

e.g.
Analysis of data for vacuum cleaner planning

Analysis for vacuum cleaner planning: [Business Data, Social Media]→[Sentiment Analysis]→[Positive, Negative, Neutral]→[Negative: noise, storage, stock, color, dust, stand, air]→[How to empty dust container:]→[Identify Needs]→[Implement product planning to improve the usability]→[Increase sales and reduce lost opportunities]

02

Implementation of prompt measures against risks, such as social media flaming and recall

category :
  • Public Relations
  • Publicity
  • Sales
  • Quality Assurance
  • Call Center

Perform sentiment analysis of data that includes both business data (such as product specifications and information about the state of business when the product was released) and information from social media. This enables you to detect negative feelings about the product before they spread, and to identify words that indicate risks. This helps you to improve your brand image by promptly implementing measures against the risk, such as preventing accidents that may be caused by defects in the product or announcing a recall.

e.g.
Implementing measures against risks related to a washer/dryer

Measures against risks related to a washing machine

03

Effective promotional activities

category :
  • Public Relations
  • Publicity
  • Sales

Perform sentiment analysis of data to monitor customers opinion from social media. Based on the customers opinion about your promotion, you can make an effective improvement for your promotion. Effective promotion activities increases your social media account followers in your business. This will make your account more attractive.

e.g.
Improving the promotion activities against customers' negative opinions

Improving the promotion activities against customers' negative opinions

04

Detection and fast handling of flaming incidents
(Sentiment Analysis × Moral Analysis × Insight Analysis)

category :
  • Public Relations
  • Publicity
  • Sales
  • Planning
  • Quality Assurance
  • Call Center

By using the Sentiment Analysis and Moral Analysis functions to analyze response on social media and the web, you can swiftly detect signs of flaming. You can then identify the causes of flaming by gaining a quantitative understanding of the moral values that led to the flaming. Furthermore, the Insight Analysis function can be used to detect potential causes for flaming before an incident occurs and to check the minority response. By using these three types of analyses, you can address risks swiftly and properly, thereby maintaining and improving your corporate brand.

e.g.
Understand negative responses as well as their background and causes, and implement swift risk countermeasures

Understand negative responses as well as their background and causes, and implement swift risk countermeasures: [Information from Social media, Web]→[Sentiment Analysis]→[Risk detected!] (Sudden increase in negative sentiments)→[Delve into the background] (There are various reasons for the negative sentiments)→[Moral Analysis]→[Many items in the "Fairness Vice" category]→[It becomes clear that negative sentiments increased because the "Fairness" value was violated, generating a lot of criticism for discriminatory speech/actions.]→[Insight Analysis] Minority opinions related to "Care" discoverd.→Based on the minority response, identify the risk's origin and changes in response.→[Detect controversial issues that might lead to flaming. Understand both the majority and minority response.]→[Promptly address the associated risks]→[Maintain and improve your corporate brand]

 *
The specifications of the service are subject to change without notice for improvement.
 *
Twitter is a trademark or registered trademark of Twitter, Inc. in the United States or other countries.
 *
Other described company names and product names are registered trademarks or trademarks of their respective owners.