编组
ConsoleUser Center
Solution Introduction

Assisting Investment Decision-making Through Fine-grained Public Opinion Analysis

Public opinion analysis plays a very important role in many business links in the financial industry. For example: rapid perception of unexpected negative events, assistance in corporate credit rating, quantitative transactions, etc. Based on the Mengzi lightweight pre-training model technology, guided by the business logic of public opinion analysis in the asset management industry, we created a financial public opinion solution by applying a number of cutting-edge natural language processing technologies.

Business Scenarios and Pain Points

Pain Points

01

page.solution.sentiment.scene.title1

Low Accuracy Rate of Sentiment Judgment Based on Keywords

Traditional algorithms judge the sentiment of articles based on keywords and rules, it can not consider word segmentation and context, and lacks the ability to understand semantics.

Pain Points

02

page.solution.sentiment.scene.title2

Emotional Judgments Do Not Align with Business Needs

The public opinion analysis data provided by existing service providers generally suffers from problems such as inaccurate judgment and insufficient negative discrimination. Moreover, direct business-oriented demand feedback is also difficult to optimize and adjust in a targeted manner.

Pain Points

03

page.solution.sentiment.scene.title4

Lots of Repetitive Information Creates Noise

In order to perceive important public opinion in the first place, enterprises often need to access multiple information sources at the same time. There are a large number of mutually reprinted articles among various information sources, which brings a heavy burden to business personnel to monitor public opinion.

Pain Points

04

page.solution.sentiment.scene.title3

Objects That Cannot Be Identified

Traditional public opinion analysis techniques can only use keywords and rules to judge sentiment at the text or sentence level, it can not make accurate judgments when multiple target companies are mentioned at the same time.

Financial Public Opinion Analysis Solution

Emotion Understanding Based on Semantics and Domain Knowledge

A true understanding of rich and varied language and domain expertise

Based on the Mengzi pre-training model technology system, text emotions can be understood from a semantic perspective combined with context. Through the study of massive professional basic corpus in the financial field and the judgments of front-line public opinion analysts, the accuracy and professionalism of public opinion judgments have been greatly improved.

page.solution.sentiment.solution.method1.title

Multi-granularity and Comprehensive Analysis of Public Opinion

Comprehensive analytical capabilities from chapters, paragraphs, to corporate entities

Using the context understanding ability and abstraction ability of Mengzi pre-trained language model technology, the target article can be analyzed from multiple angles. Cooperating with corporate entity identification and chain index, it can accurately identify and correlate tens of thousands of listed bond-issuing companies, and accurately locate the different emotions for each corporate entity in the intricate description.

page.solution.sentiment.solution.method2.title

Massive Article Intelligent Deduplication

Efficiently locate important public sentiments in massive data

Through the Mengzi pre-training language model technology, it is possible to judge whether the description content is repeated based on the abstract understanding of the content of the article, and effectively identify the slightly rewritten repetitive articles that are difficult to distinguish using rules. It prevents business personnel from being overwhelmed by a large number of repeated messages, thus unable to pay attention to important public opinions on time.

page.solution.sentiment.solution.method3.title

Business Value

page.solution.sentiment.value.title1

Effectively Improve the Accuracy of Negative Public Opinion Identification and Classification

In terms of the accuracy of public opinion classification, the engine helps each downstream business system to accurately identify multi-level public opinion information, effectively improves the distinction between public opinion at each level, and provides more fine-grained decision-making support signals for the business side.

page.solution.sentiment.value.title2

Filtering Articles from Companies that are not the Main Target of Public Opinion

The main body of the company concerned by business personnel is often mentioned in a large number of articles, but most of the time it does not appear as the main object of public opinion. By analyzing the description content of the full text, whether it is the main object of public opinion can be judged, and a sentiment analysis for this company can be provided.

page.solution.sentiment.value.title3

Reducing Time Spent Manually Processing Repetitive Information

With the development of big data technology, the number of public opinion information sources accessed by financial institutions has reached tens of thousands. Through massive information deduplication technology, articles with repetitive content can be largely filtered to reduce noise interference by business personnel.

Application Cases

China AMC

The public opinion system realizes processing capabilities such as public opinion subject extraction, subject-oriented emotion recognition, and semantic-based article content deduplication, and can perform multi-level sentiment judgments on public opinion subjects, quickly locate the company subject of concern from massive public opinion information. The positive and negative information is stored in a unified data platform and provided to multiple downstream business systems. At the same time, through the pre-training model domain adaptation and model lightweight technology, a large amount of computing resources are saved under the premise of ensuring excellent results.

https://cdn.langboat.com/portal/page.solution.sentiment.case1.title

Solution Experts Can Demonstrate Products Remotely or on Site

Related Products

Langboat Meeting Assistant Platform

The Langboat Meeting Assistant can efficiently realize the functions of speech to text and multi-dimensional intelligent analysis of conference. Applicable to office meetings, teaching speeches, media interviews and other meeting scenarios,Provide deep analysis and value mining of meeting contents.

Financial Applications

Providing various NLP capabilities with strong versatility in the financial industry through APIs.

Products

About Us

Mengzi Community

Great News

Join Us

Business Cooperation Email

Address

Floor 16, Fangzheng International Building, No. 52 Beisihuan West Road, Haidian District, Beijing, China.


© 2023, Langboat Co., Limited. All rights reserved.


Large Model Registration Code:Beijing-MengZiGPT-20231205


Business Cooperation:

bd@langboat.com

Address:

Floor 16, Fangzheng International Building, No. 52 Beisihuan West Road, Haidian District, Beijing, China.

Official Accounts:

ewm

© 2023, Langboat Co., Limited. All rights reserved.

Large Model Registration Code:Beijing-MengZiGPT-20231205

support
business