<h2 dir="ltr" style="text-align:center">What is natural language processing, and how different is it in a financial context? 
&nbsp;</h2>

Natural language processing is a form of artificial intelligence that deals with the interaction between computers and humans to better understand, interpret and manipulate human (natural) languages. NLP takes on many forms but the ultimate goal of NLP is the cognition, understanding and generation of human languages in a manner that is valuable.

&nbsp;

In order for machines to derive meaning from natural language, the computer will utilize algorithms to turn textual input into associations and relationships extracted within a sentence, and from there collect and process according to a purpose.

&nbsp;

NLP is the driver behind many forms of AI you&#39;re used to seeing. Whether it be translation (Google Translate), speech recognition (Apple&rsquo;s Siri, Amazon&rsquo;s Alexa), knowledge graph and intelligent graph queries (Google Search Engine) etc., NLP is enabling technologies to understand and process human languages.

&nbsp;

Natural language processing for the finance sector is still in the exploratory phase. Since finance itself is a highly specialized field, many words carry nuanced meanings in a financial context. From jargon to technical terms, the financial field measures, understands and processes language differently from other industries.

&nbsp;

<img src="https://lh6.googleusercontent.com/nGnDSjiQpceRMm0fV8FcZpgAnQ-2Fuj4N98jCVCnqrpMLtaee_c5Mp9r4NnTKA5vVu5nShTIcQxKD1lP67IxUG8tJaV9Vs63VVB-83XfWdLbZ3aazVz7S60NnZGZ1FGiA0Bwk_dS" style="height:423px; width:595px" /> 
Classification and scoring of news sentiment in AMI, MioTech&rsquo;s market intelligence tool.

&nbsp;

Therefore, NLP to serve the financial sector requires the preparation of a specialized training data set. At present, all NLP architecture requires large data sets which leads us to the biggest challenge NLP faces in the financial field, the lack of a deep, professional training data within finance.

&nbsp;

At MioTech we apply NLP to focus on solving NER, relationship extraction and knowledge graph. Using knowledge graph to compensate for the inadequacy of training data, the data set can be supplemented with other data already associated within the map.

&nbsp;

<img src="https://lh3.googleusercontent.com/j5gsgfQSo8FSP4BHJjVx_8OpZ_EY9VVtTqKpEG84dzFK_Pi0b7nHrEb88rPJJ0cwnDdVgjJe0pA7EoQUsLaTNnB0oyaN-ZYPL_W_dzPOO-jcTGi1kWnvHVCctjntbOkB9oA4ciuB" style="height:396px; width:624px" /> 
AMI&rsquo;s knowledge graph &nbsp;

&nbsp;

<h2 dir="ltr" style="text-align:center">Where are we at with NLP and how difficult is this process?</h2>

&nbsp;

The development process of natural language processing is roughly the same as the development of artificial intelligence, and has experienced the following stages of development:

&nbsp;

<ul dir="ltr">
	<li>The 1950s-80s: Realization of human mastery based on human experience;</li>
	<li>From the 1990s to around 2000: Based on statistical methods;</li>
	<li>From 2000 till now: Due to the substantial increase in data and computing power, the gradual introduction of the deep learning method into the NLP field saw major breakthroughs in machine translation, question and answering systems, and automatic summarization. 
	&nbsp;</li>
</ul>

But it is important to note that Natural Language Processing still faces many challenges. The human language is very concise and omits presumed knowledge in many conversations. Humans have the ability to easily understand the context even with omissions. But for a machine, it&rsquo;s a complex process. Take for example, &ldquo;I&rsquo;m engaged.&rdquo; When a machine doesn&rsquo;t understand the specific context, it is difficult to distinguish whether the person is engaged to another person or if the person is temporarily unavailable. As you can see, parsing English with a computer is not an easy task.

&nbsp;

<h2 dir="ltr" style="text-align:center">How different is NLP in English as compared to Chinese? What are the challenges facing NLP Chinese as a language? What are the best methods&nbsp;for training Chinese NLP?</h2>

&nbsp;

From a pure language point of view, English comparatively to Chinese, is more direct. By extracting just the nouns within a sentence, you can infer the lexical semantics to a large extent. Since English is a phonetic based language, one can also infer intention by parsing through grammar, tenses, passive/active, affixes, singular/plural forms of words.

&nbsp;

But Chinese language or characters are similar to hieroglyphs, meaning that there is limited conversion on various parts of speech and it is near impossible to split a single word. Therefore, the machine must judge the specific semantics through context. Due to the uniqueness of Chinese language, the performance of same NLP model in the English context is generally better than in Chinese.

&nbsp;

Word segmentation is one of the difficulties in Chinese NLP. For an English comparison, take for example, &ldquo;A woman without her man is nothing&rdquo; or if split/punctuated, &ldquo;A woman/ without her/ man is nothing.&rdquo; Different word segmentation causes ambiguity.

&nbsp;

With the widespread use of deep learning, what sets different languages apart have gradually become a difference in the amount of training data. In the past, in terms of the NLP field, the amount of Chinese data available was far less than that of English data. However, now in China, more and more people are investing in artificial intelligence and research in the field of NLP. Therefore, the problem of insufficient Chinese data sets is improving year on year.

&nbsp;

In the financial field, the stages of Chinese and English NLP are generally the same in terms of basic financial knowledge. But, for complex issues such as sentiment analysis, relationship extraction or topic extraction, both Chinese and English NLP have long way to go due to the complex scenarios and context, as well as the substantial amount of training data.

&nbsp;

<h2 dir="ltr" style="text-align:center"> 
What do financial institutions seek to gain from a robust NLP solution?</h2>

&nbsp;

NLP can aid with analyzing an industry&#39;s entire value chain and sentiment analysis including monitoring and analyzing news, announcements and social media. Machines can help us read and process huge amount of information in a short time which is impossible for human beings to do so.

&nbsp;

For example, commercial banks want to utilize comprehensive data to conduct accurate credit risk management of enterprises and to predict or forecast the potential risks posed by these enterprises in advance. At present, the conventional method would be based on the annual report published by the company and then judged according to the results from the field investigation conducted by the compliance officer. The enterprise&#39;s own risk disclosure not only comes with a time lag between the report and the actual event, public information coverage is not high, resulting in an assessment that covers only a basic understanding or just the tip of the iceberg. This is where natural language processing and artificial intelligence can make a difference.

&nbsp;

Natural language processing can mine the multi-dimensional relationships extracted within information, evaluate the relationship between enterprises, and intuitively present the relationship between enterprises through knowledge graph, and from there, set up early warning alerts in advance. Once a relevant/significant change within the enterprise network is detected, the quick assessment of the impact on the entire network can be achieved based on the weight of the relationship.

&nbsp;

<img src="https://lh4.googleusercontent.com/jwmg0SVlfafFhyKkUy7L_GlLz2oLEe88j8voQcbWVCV3uZ_vlk_xGReXiELTVQDa4qQNxs0HvmrvOI2E9YxRkAMp_Bz38G-SFoA4ONeZ6XRF_-aiav6wbdYOubrU5uNMCS5HzWHl" style="height:272px; width:624px" />

&nbsp;

Mining an industry&#39;s value chain based on public financial reports of listed companies is another application of natural language processing within a financial context.

&nbsp;

Data of an industry&rsquo;s value chain are based on the financial reports of A-share listed companies, all of which are original data sources. According to the main business composition in public financial reports, the keywords are extracted and input into the pre-trained neural network, and the vector expression is aligned.

&nbsp;

Next, we perform density-based clustering on the input vector, output clusters of different densities, and finally cluster naming.

<img src="https://lh3.googleusercontent.com/1SYGpe2bnrkXtksFgkgxUDVUDHRCcV75lXo6MgBMemfAd0YTjANYj5zr-M2i5p5yUDeyJ3BE7KJYe-pKeIhTnGt5M_fbLEcRdqs6b3kmt-q76k7qovdU-8q6ZABDGTaVyYsLiJ-_" style="height:264px; width:624px" />

<img src="https://lh5.googleusercontent.com/S78XIpHU75cwhkZVhZ_FwLmgT56KnhAMJ44eOcjka9g0eeBn9ED12KS8_j0WHlQCkJ_7svTGEM9Px0nTOc-DpKdArIJr_aVOMx8ewwz49JHI0woFQFLT3eH_DI8mhMNVWLGHIplb" style="height:391px; width:624px" />

&nbsp;

<h2 dir="ltr" style="text-align:center">What do you foresee happening in the future in terms of NLP, especially for Chinese?</h2>

&nbsp;

As more and more data is generated each day, the number of data sets available for machine training will continue to increase. With this continued generation of data, NLP can only increasingly see significant improvements.

&nbsp;

At the same time, with the development of deep learning, machines can now solve targeted problems more directly without relying on human input. This can be seen in AlphaGo which continued to perform better and then outperform after absorbing human experience.

&nbsp;

After the emergence of the BERT model, the accuracy of NLP has improved. In some cases, take machine reading comprehension in particular, the accuracy of some models have completely surpassed humans. This year, AI beats humans in a Stanford reading comprehension test - Stanford Question Answering Dataset (SQuAD). The leaderboard on <a href="https://rajpurkar.github.io/SQuAD-explorer/">SQuAD&#39;s website</a> shows machine&rsquo;s EM score (Exact Match, providing exact answers to questions) to be 87.147 (as of Mar 20), which puts it at first place, using the BERT algorithm. The score is higher than a human&#39;s, which is at 86.831.

&nbsp;

From the perspective of the Chinese language, NLP will continue to develop in tandem with deep learning. As data sets become more and more abundant, the relationship extraction for complex semantics will be more accurate, and the recognition of intentions will gradually improve. In specific application scenarios, especially in the financial sector, it will grow to become more mature.

&nbsp;

At MioTech, we&rsquo;ve adopted the knowledge map to solve the problem of insufficient training data. With the further development of NLP algorithms, paired with our core technology, entity recognition and relationship extraction will only be more accurate. 
&nbsp;

In this article, we speak to MioTech's CTO on the fundamentals of Chinese NLP in a financial context, its challenges, its progress and what's in store for this form of artificial intelligence in the near future.

Chinese NLP 101

Co-Founder & CTO, MioTech

<h2 dir="ltr" style="text-align:center">What is business intelligence?</h2>

In the course of developing a business, it&rsquo;s quite likely that you&rsquo;ll encounter certain decisions. It&rsquo;s fundamental to ask yourself if these choices are influenced by personal experience, based on logic or is it based off a more well conceived plan? The role of business intelligence is to eliminate guesswork and complexity by harnessing data to make more informed business decisions. Data-driven business outcomes reduces postulation, bias and rather provides solutions based on solid theories, methodologies, processes, architectures, and technologies.

&nbsp;

<h2 dir="ltr" style="text-align:center">How can business intelligence be achieved? 
&nbsp;</h2>

Business intelligence within a business, specifically tech companies, typically includes the following components: 
&nbsp;

<ul>
	<li dir="ltr">
	Data Mining and Analytics: By adapting data mining tools and integrating business datasets to streamline the analysis process. The common form of presenting business intelligence is through data visualization for extracting insights and greater verification. 
	&nbsp;
	</li>
	<li dir="ltr">
	Anomaly Detection: &nbsp;By using algorithmic machine-learning methods to identify aberrations or even minor blips in expected patterns in a virtually unlimited number of metrics in real time. For example, on a product level, it can identify the number of DAU, detect customer retention and financial risk control. On a business level, it can chart employee turnover and financial metrics. 
	&nbsp;
	</li>
	<li dir="ltr">
	Data Services: Focusing on data output, self-service tools to perform queries for each department. Meta-data and rights management, forming data middle office, supporting the data needs of other products.
	</li>
</ul>

&nbsp;

<h2 dir="ltr" style="text-align:center">What are the daily tasks of data analysts/scientists?</h2>

To achieve &ldquo;data-driven&rdquo;, intelligent decision making. Effective data is the cornerstone of analytical mining.

&nbsp;

Where does the data come from?

Incomplete statistics, a professional working in the data field spends 70% of his time on obtaining data. Depending on individual business needs and the type of product, we require all various kinds of datasets in order to obtain raw data.

&nbsp;

<ul>
	<li dir="ltr">
	Data is usually generated internally within the business. For example, an e-commerce platform needs to tally the order, quantity and distribution of the commodity. Food delivery services need daily statistics on their orders, quantity, timings. Ride-hailing services generate data just off every single ride. Generally, all this data would be stored within the product universe. In order to optimize your analysis, you&rsquo;d need to integrate your internal, real-time data with synchronization tools.
	</li>
</ul>

&nbsp;

<ul>
	<li dir="ltr">
	With regards to user generated data, the behavioral trajectory of users in products often provides intuitive feedback on product decisions. Take for example a news platform, you can track the page views of popular news, or a heat map of where users click on specific webpage. Users inevitably leave a series of &quot;footprints&quot; in the process of use. Generally, you can obtain user behaviors through front-end management, back-end logging, etc. Hence, it is essential to accurately locate and detect user movements and be disciplined about recording it. Take payments for example, from the time an order is submitted to securing a successful payment, there are a number of points within that process where users can choose not to proceed. But this gives you a data point. Based on the behavior, you can draw user profile/persona.
	</li>
</ul>

&nbsp;

<ul>
	<li dir="ltr">
	In terms of access to third party data, this can be obtained through data providers such as ad-click through rates etc. In addition, when data analysis has output capabilities, you can seize the opportunity to work with enterprises that lack independent analysis capabilities and complete analysis work for them.
	</li>
</ul>

&nbsp;

How is the data analyzed?

After we have obtained the necessary data for analysis, completed data preprocessing, model construction, and warehousing into a unified data warehouse, how should a business tackle analyzing this data?

&nbsp;

Analysis of baseline

Different departments within an enterprise usually have different indicators of their concerned about. The market may be concerned about channel conversion rate. Sales may be concerned about payment. Personnel may care more about success rate, turnover rate and so on. The first step for analysts is to provide key metrics that clearly reflect the current state of the company or the current state of the product and present it to decision makers in a timely manner.

&nbsp;

Drill into Root Cause 
 
Identifying the baseline of the indicators is still insufficient to deliver meaningful analysis or insights. We also need to know how the baseline was formed. Analysts need to review the entire process with decision makers, chart the full life-cycle of the data, and know why the baseline is the way it is.

&nbsp;

For example, if the success rate of recruitment is relatively low, we should review the entire process and perform data funnel statistics on every stage, from resume screening, telephone interviews, face-to-face interviews and finally, entry. If the pass rate of resume screening is relatively high, but the success rate of telephone interviews is low, then it is very likely that the screening of resumes is too loose. This is insights that we can provide to HR.

&nbsp;

For example, if we look at product user retention rate, if the 30-day retention rate is relatively low and the user loss is serious, then the user activity can be observed weekly, monthly for loss and retention rates. If the user&#39;s 7-day loss rate is relatively large, it is necessary to act fast at the corresponding time node to sustain active use of the product.

&nbsp;

Forecast

In hindsight, we should also be forecasting ahead. Based on historical records, we can also predict possible trends so that decision makers can make sharper choices and from there, analysts can estimate the possible outcomes and prepare for it. This requires us to use statistical methods, machine learning and other complex algorithms to learn historical data and derive estimates.

&nbsp;

For example, it was estimated that the annual volume of data collected from TMalls&rsquo; Single&rsquo;s Day sales had the ability to predict customer fraud, and it predicted the expected loss of customers in advance.

&nbsp;

Predicting the future helps decision makers understand risks ahead of time, seize opportunities, and get the best results.

&nbsp;

<h2 dir="ltr" style="text-align:center">How should one look at the data?</h2>

 
The deliverables for data analysis are typically statistical results that are aggregated in a data form. Through data visualization tools like MioTech, we can visually present data characteristics and patterns in multiple dimensions. In every report, it is essential that we present macro-statistics that have the ability to not only give an overview, but also drill down to a granular level within its dimension. 
&nbsp;

<img src="https://lh5.googleusercontent.com/U7_lzc0D-Wfk2HP2dOGL9th0J2euw4NFdTGW_exEVXTunoywou7Iqh3e4pob1JgL64rupFeqYuweqHommO1QLe7GQscjXJrXItOczLajbBiYqvqkOjY0NQ_CBXD7nCzkNN4JJQed" style="height:249px; width:624px" /> 
BAT (Baidu, Alibaba, Tencent) investment details from 2009 to 2019, source: MioTech AMI

 
&nbsp;

<h2 dir="ltr" style="text-align:center">Why is business intelligence and data analysis increasingly important in the financial sector?</h2>

&nbsp;

The types of data services

In addition to empowering internal decision making, data analysis provides data support for the broader user community. Let&rsquo;s use the tech industry as an example to discuss a few types of data analysis services.

&nbsp;

<ul>
	<li dir="ltr">
	Analysis based on user data
	</li>
</ul>

Users create its own data history through in-app transactions, content creation and other browsing and clicking activities. Through multidimensional analysis and statistical analysis of historical data, companies can gain user characteristics and behavior insights. For example, Meituan or the Ele.me provide merchant services, WeChat for Work provides work summaries, Alipay provides annual statements, etc.

&nbsp;

<ul>
	<li dir="ltr">
	Custom Analysis
	</li>
</ul>

When users are not satisfied with the fixed statistical analysis dimensions provided by the product, they can use the analysis service by providing a custom query. For example, Jira&#39;s JQL custom query allows high-level users to complete queries using the semantics of class SQL. Even with internal analysis resources, you can provide analysis services for specific customers, meet their additional needs and deliver them through periodic mail or reports.

&nbsp;

<ul>
	<li dir="ltr">
	Data Analysis Service providers like SaaS
	</li>
</ul>

They do not generate data themselves, but can provide standard solutions for specific industries; like data access, data processing, and analysis of the entire data stream. For example, Google Analytics can provide general website traffic analysis.Growingio and Sensors Data,successful Chinese companies, focus on industry analysis scenarios such as Internet e-commerce. Users can choose to submit data to public network platforms or privatized services.

 
&nbsp;

<h2 dir="ltr" style="text-align:center">Data Analysts or Data Scientists? What&rsquo;s the&nbsp;difference?</h2>

 
There may be a distinction between data analysts/data scientists when recruiting in larger companies, but in smaller companies the boundaries between the two are rather vague. There is a saying which I think is more accurate. These two job roles reflect the broader internal shifts towards specialization within traditional service enterprises as they enter into the big data era.

&nbsp;

The skillsets you need to become a data scientist:

<img alt="Related image" src="https://devopedia.org/images/article/38/2255.1523605917.png" style="height:569px; width:500px" /> 
&nbsp;

&nbsp;

For traditional data analyst, we usually pay more attention to his industry knowledge, business knowledge and analytical skills. He needs to have a wealth of industry experience, can quickly locate the required indicators according to the needs of the department, skillfully using tools to produce the corresponding chart or insight.

&nbsp;

In the era of big data, the magnitude of data has increased exponentially on the macro level, and the types of data have become complicated. This requires data workers to use big data tools to fully capture data from multiple data sources, consume massive amounts of data, and ensure stable output.

&nbsp;

For each piece of data, we want to extract as much information as possible. Therefore, we can&rsquo;t be satisfied with only counting the number of commodity reviews.We need to also conduct sentiment analysis, keyword extraction, and understand user feedback. Current statistics alone isn&rsquo;t enough for us. &nbsp;We don&#39;t just need to know what users historically like the most, we also need to speculate what users might like in the future. It&rsquo;s insufficient to soly analyze structured data, unstructured data like images, audio, video and geographic data are also essential objects of analysis.

&nbsp;

Therefore, more skills are required of data workers today. But learning technical skills is a relatively easy process, especially in the era of big data. The emergence of various automated tools provides more convenience for data processing. While data sensitivity, industry understanding and product understanding takes more time to digest, they are still more valuable than skills.

In this article, we speak to MioTech's Senior Software Engineer on the fundamentals of business intelligence, why it's indispensable for business growth and how it can enable better decision making.

Business Intelligence 101

Senior Software Engineer in Data Infrastructure

<blockquote>
<h3 dir="ltr" style="text-align:center">What is a knowledge graph?</h3>
</blockquote>

&nbsp;

In simple terms, a knowledge graph is a tool that visually represents all entities and their interrelations/relationships in a knowledge domain in the form of graphs in a network structure. Common entities include: company, person, place name, and an event. Common relationships include: competitive relationships, foreign investment, and so on.

&nbsp;

<img src="https://lh3.googleusercontent.com/osgnGEw-MuuGtTuFqCHnqVCMGa9ymSoTV7c6XHBCmvcB5PPadsHL3OuPUlKIjQfzJc6r3nX3ViR_gR7Lvic_IW4J_UfvD2UO_2Loy5SZYUYRoWxxzQTU1tzcbww3sDqujvZcT1Wl" style="height:304px; width:540px" /> 
Source: MioTech Report

 
&nbsp;

<blockquote>
<h3 dir="ltr" style="text-align:center">Before a search engines started using knowledge graph, there was the &ldquo;search string&rdquo;, could you explain? Why was that inefficient?</h3>
</blockquote>

&nbsp;

Before the emergence of knowledge graph, search engines only used text data string entered by the user to match the text string on all the web pages to find the relevant information.

&nbsp;

For example, let&rsquo;s say I wanted to know when &ldquo;Captain Marvel&rdquo; is playing in the cinemas.&nbsp;I would have had to first, search for the cinema, then scroll through the website to find the right movie, and then find the showtimes. I would have had to navigate through multiple links before I finally get&nbsp;to what I initially wanted.

&nbsp;

This simple way of searching information has its advantages in being able to find out exactly what you want quickly but its limitations are obvious. First of all, this practice of merely matching text does not fully understand the user&#39;s true intentions which often &nbsp;results in unanswered, unfulfilled search queries. Secondly, simple string matching cannot extend the information further, which greatly reduces search efficiency.

 
<img src="https://lh6.googleusercontent.com/NNpVdzAEsmO_1TJqpl0j5_Mo6Uugsd9utucVOoxQMsNphoniejkNjVIvfpvTMe1dEDQ5uC6rJde7o-uML4PZny-gaoYX8oBaiUnfjeU4JvNbIqHm5cUik0jU1utOayNJ0J5FIWrn" style="height:304px; width:540px" /> 
Source: Company Report 
 
&nbsp;

<blockquote>
<h3 dir="ltr" style="text-align:center">Why is the knowledge graph more efficient? How is the AI infrastructure better as compared to others?</h3>
</blockquote>

&nbsp;

Knowledge graph is the knowledge base behind AI. Without the support of knowledge graph, the amount of information that the AI ​​can receive is insufficient. The two are interdependent.

&nbsp;

Let&#39;s go&nbsp;back to the previous example of wanting to know when &ldquo;Captain Marvel&rdquo; is playing in the cinemas.&nbsp;With today&rsquo;s knowledge graph structure, I will not only get an array of cinema choices and movie timings, it&rsquo;d be paired with the movie&rsquo;s official website, perhaps a Rotten Tomatoes score, perhaps a news article in relation to gender representation in Hollywood. Moreover, depending on my user behaviour, it&rsquo;ll rank the information based on what I would like to click based on previous searches.

&nbsp;

The cornerstone of AMI, MioTech&rsquo;s virtual data scientist&rsquo;s graph query, is our proprietary knowledge graph. If users want to know Apple&#39;s stock price, with the help of AMI&rsquo;s knowledge graph, our graph query can present all information related to Apple, such as financial reports, competitors, foreign investment activities, latest news updates, etc. to users,&nbsp;significantly improving search efficiency for a financial context.

&nbsp;

<blockquote>
<h3 dir="ltr" style="text-align:center">How does a knowledge graph help AI discover hidden relationships? 
&nbsp;</h3>
</blockquote>

Two examples can be given to illustrate this. First, in the field of anti-fraud, through knowledge graph, we can visualize certain patterns from historical fraudulent activity/data, and evaluate whether other multiple transactions satisfy or share a similar pattern and if so trigger the fraud warning to alert users.

&nbsp;

<img src="https://lh6.googleusercontent.com/ved9FwtgvfNbZ4lts5WfmNy9CTK6RgOlQIwVElOE218go34xsniLL735vj4W_Nm_DXl4ddzQbmBxOHzgoKo4K7cbaUAbUupZugHGV1HOJVuAd7v7rAogrvkPYp_o57_VgelK4tcC" style="height:304px; width:540px" /> 
Source: MioTech Report 
&nbsp;

&nbsp;

Another common example is when conducting market research. A merger of a company taking place in Los Angeles may have a considerable impact on London competitors, Chinese suppliers, and Taiwanese competitors of this Chinese supplier. With the help of knowledge graph, we can consider these isolated events as a whole.

&nbsp;

<img src="https://lh4.googleusercontent.com/Tjie0OSSdWtJhr49SRR1yYCmWTI5Wpe0M0SQfrB3NqUikkOFgYVKR9yPx9DWqGRXUrRP0mZlAF88GWnKjEMe_w-UZQxrbJniRdE62CfoovWjLii2HoTReR5L4GQ0VEEHK7GL08l7" style="height:305px; width:540px" /> 
Source: MioTech Report

 
&nbsp;

<blockquote>
<h3 dir="ltr" style="text-align:center">How is MioTech&rsquo;s knowledge graph different from Google?</h3>
</blockquote>

&nbsp;

From a data perspective, Google&#39;s knowledge map can be called the &quot;General knowledge graph&quot;. This means that Google&#39;s data is more comprehensive with data coverage being broader, catered to the wider general audience. But MioTech&#39;s knowledge map is tailored to the financial industry. All the data in our database is financially related. In terms of coverage of financial data, we can say that we have more data coverage than Google. 
 
At the application level, MioTech&rsquo;s knowledge map is more in-depth. We cater and customize to the different needs of the financial industry, such as equity penetration analysis, peer analysis. These are results that Google&#39;s knowledge map cannot achieve.

&nbsp;

<blockquote>
<h3 dir="ltr" style="text-align:center">What is critical when building MioTech&rsquo;s knowledge graph? 
&nbsp;</h3>
</blockquote>

Let&rsquo;s compare it with Google. From a data perspective, Google requires more complete data, but there is no higher requirement for the reliability of the data source. In the financial industry, there is a higher demand for accuracy, reliability, and real-time data, because it is directly linked to investment decisions.

&nbsp;

<blockquote>
<h3 dir="ltr" style="text-align:center">What are the challenges knowledge graphs currently face in terms of performance. How can it be improved?</h3>
</blockquote>

&nbsp;

The reliability and real-time nature of data sources is currently a problem we need to solve. 
 
In order to ensure the reliability of the data, we have designed a data testing framework. Common data tests use a method of extracting certain samples. But in order to ensure the high accuracy of data in the financial industry, we have performed a full data scan. The disadvantage of this method is that it is slow. But we are passing distributed systems and parallel computing to increase the speed of detection. 
 
The real-time nature of the data is another challenge we are currently facing. At present, most of the data used by knowledge graph comes from traditional data providers and the method of manually entering data is still used. Therefore, new events often take a while to be reflected in the knowledge map. We are currently developing a new AI model to combat this. After completion, we will no longer rely on labor, but let the machine determine the subject, event and object; and such combination will result in new relationships in the knowledge graph, resulting in more hidden market signals or investment opportunities to be found. This AI model will take into account the source of the message, the number of media mentions and related contexts to ensure that the data, while real-time, is accurate.

 
Training labels are also a point we need to pay more attention to. The financial industry has higher requirements for the labeling of labels, and has certain requirements for the qualifications and financial background of the labeling personnel which would come at a substantial cost.

&nbsp;

<blockquote>
<h3 dir="ltr" style="text-align:center">What is in store for MioTech&rsquo;s knowledge graph?</h3>
</blockquote>

&nbsp;

In the future, MioTech&rsquo;s knowledge graph will no longer need manual labeling.&nbsp;With the AI ​​model I just mentioned, automatic labeling can be achieved.

In this Q&A article, MioTech’s Director of Data Infrastructure and Science, Alex Zhou gives us the low-down on the knowledge graph and why they play a huge role in many intelligent systems today.

The Source of all Knowledge

Director of Infra and Science

Colloque Tsui is MioTech&rsquo;s senior front-end engineer, responsible for creating MioTech&rsquo;s core technology, AMI. Just 6 months into joining the tech squad, Colloque, or Cola as we MioTechers endearingly call him, dived in head first to develop AMI&rsquo;s Knowledge Graph. &ldquo;At MioTech, new challenges will come at you anytime. There&rsquo;s something new everyday.&rdquo; Originally from Hong Kong but based in Shanghai, Cola spends roughly 50% of his time developing new features, 20% troubleshooting issues, 10% managing and updating the ever-changing task managing system, Asana, and the last 10%, fighting for an alternate career as a Street Fighter Esports player.

&nbsp;<img src="http://miotech-landingpage-images.oss-cn-shanghai.aliyuncs.com/images/1550887713245_IMG_1760.jpg" />

<blockquote>
Why did you decide to move to Shanghai and take up a job with MioTech?
</blockquote>

As a software engineer, the opportunities for good developer roles are better in the Mainland as compared to Hong Kong. Upon graduating from the UK and returning to Hong Kong, I realised that there weren&rsquo;t many reputable IT companies locally that I was keen on joining. The companies that I had worked for in the city were small in size and the projects involved geared towards e-commerce and online marketing/advertising. While it had the flexibility to apply technology to an actual application within a short period of time, I was dabbling in smaller, seasoned software. Personally I felt there wasn&rsquo;t enough room to grow which is why I ended up leaving.

<img src="http://miotech-landingpage-images.oss-cn-shanghai.aliyuncs.com/images/1551406289737_image2.png/th1000" style="height:auto; max-width:100%; width:auto" />

But if I&rsquo;m completely honest I moved to Shanghai for love. I chose MioTech for its youthful, energetic and entrepreneurial spirit. (MioTech&rsquo;s average age is 28.8 years old) It became clear to me 6 months into the job that it was the right choice. On top of growing in my career, I&rsquo;m growing a wider Mandarin vocabulary. MioTech&rsquo;s well stocked pantry also means a wider waistline.

<blockquote>
What is the biggest difference between MioTech and your previous job?
</blockquote>

The biggest difference between my previous work experience and MioTech is the level of access and exposure I get to the most cutting-edge technology in the industry. This privilege comes with it the need to learn continuously and be in synch with the latest offerings in the field.

In order to expand the capabilities of of our knowledge map, the tech team decided to rewrite the entire code to convert from SVG to Canvas. While the deadline was tight, it was a moment for me to test myself. Not only has the FPS greatly improved, our additional adjustments to our recently launched Shareholding Analysis function gives us that edge against our competitors.

The other big change was moving from a B2C to a B2B business. I was responsible for web development at LIHKG, now one of the largest local forums in Hong Kong. One of the features I&rsquo;m most proud of is the &ldquo;office mode&rdquo; function on the website. A lot of our users love to camp on the website, browsing it during work time, so we decided to create an incognito office mode for them to appear like they&rsquo;re working. So what we ended up doing, was to disguise the website as an Excel or Word document so eyes from a distance would assume you were working. Positive feedback was instant. You do feel a great sense of accomplishment working in the 2C environment as feedback is instantaneous upon launch of a new feature. But it was a relatively simple product.

<img src="http://miotech-landingpage-images.oss-cn-shanghai.aliyuncs.com/images/1551405041614_image4.jpg/th1000" style="height:auto; max-width:100%; width:auto" />

After joining MioTech, it was a huge learning curve for me when it came to developing B-end products. The concept of complete product development was brand new to me. AMI was much more complex application with heavy data visualization features. This meant delving into Canvas, D3 and Highcharts fast. I had to adapt fast to sprints which I never had before in my previous job. On top of the pace of work and the rapid iteration of products, it really does push you to think more strategically and carefully on developing an architecture of this scale. But being exposed to a bigger team meant tapping on the knowledge and support of many talented people. While you do work longer and harder for gratification, I&rsquo;ve come to realise that the sense of accomplishment is just as gratifying.

<blockquote>
What&#39;s your favourite Front end stack? (ReactJS) Why? (As compared to Vue)
</blockquote>

When I first started working with modern frontend frameworks, I loved Vue. It &nbsp;was easy to use, fast and small. It still is and I still love it. But for what I do now, I prefer ReactJS. With React you have the options to build multiple platforms whether it be native mobile apps or VR apps or TV apps. The syntax of React&rsquo;s library is also a lot closer to Javascript whereas Vue has a bit of a syntactic sugar and it kind of works like magic behind the scenes if you don&rsquo;t know what you are doing. You really need to study their documentation in order to use Vue well. But I must say that Vue&rsquo;s documentation is one of the best out there.

<blockquote>
&nbsp;What are some of the challenges you face when writing code?
</blockquote>

When writing, you&rsquo;re never writing solely for yourself. It&rsquo;s critical to write it in a way that is understandable to others because you never know who might maintain your code later on. Coming into MioTech, I had to maintain the codes of my colleagues which were different from my coding style. Moreover when working on a modern front end project, you&rsquo;ll be writing modular code (components) that needs to be easily reused by others. You&rsquo;ll need to conscientiously build a structure that doesn&#39;t affect other components which could cause a range of unpredictable bugs. So I know first hand the importance of writing for the benefit of everyone. 
 
&nbsp;

<img src="http://miotech-landingpage-images.oss-cn-shanghai.aliyuncs.com/images/1551406184627_image3.png/th1000" style="height:auto; max-width:100%; width:auto" />

Source:&nbsp;Thom Holwerda 
&nbsp;

While I&rsquo;m still learning the art of it all, I try my best to write clean and understandable code. Of course there are times when a deadline is tight (sometimes tomorrow morning) and features need to be pushed out the door and my code is compromised. Even if I want to come back to it and refactor it, I&rsquo;d get snowballed with other features. But I&rsquo;m making it a point during my free time and my tasks are done, I revisit and refactor the old code.

But keeps me excited about UI engineering is this constant evolution of tools. In the frontend world, new concepts and libraries come and go. While the frontend library we use is React, we&rsquo;ve recently adopted a new concept called Hooks, which can encapsulate a chunk of code that can be reused in many components without polluting the virtual dom tree. It&rsquo;s fascinating and new, the industry is still experimenting with it and figuring out the best practices when using it. You can only grow by being part of this dynamic community.

<img src="http://miotech-landingpage-images.oss-cn-shanghai.aliyuncs.com/images/1551406203192_image1.jpg/th1000" style="height:auto; max-width:100%; width:auto" />

&nbsp;

<blockquote>
How does your day look like?
</blockquote>

8.00 Get up and browse the latest technology news

9.30 Check tasks on Asana, confirm code deployment and check on bugs in yesterday&rsquo;s new online functions

10.00 Standup meeting to confirm products that are going live. Write code.

12.00 Lunch/Street Fighter with GameSquad

14:00 - 20:00 Write code. Write more code. Write even more code. When the day ends is always uncertain.

20:00 - 24:00 Put on a movie, play games, play more games.

&nbsp;

Delighting financial institutions by making the best software he can imagine, Colloque Tsui, MioTech’s senior front-end engineer in charge of stability, performance and visual quality.

My Not So Typical Day as a Front-End Engineer

Sr. Software Engineer in Front-end

Yishu, MioTech&rsquo;s chief front-end engineer, graduated from Southeast University. He has extensive experience in Baidu and Morgan Stanley working as a software engineer. He&rsquo;s endearingly known as &ldquo;Issue&rdquo; to us MioTechers, mainly because his day revolves a lot around finding the software bugs in the system and exterminating them.

Through the daily demands of working at a fast-paced startup and an ever changing software landscape, Yishu has learnt, tested and applied the most advanced front-end frameworks in order to build up our arsenal of applications. Since MioTech specializes in data analytics and visualization through our advanced knowledge graph, Yishu&rsquo;s kept up to date with a variety of new technologies and libraries like d3, SVG and Canvas to ensure high levels of function and quality in MioTech&rsquo;s software.

Yishu has been honoured the highest MioTech employee award, &ldquo;Rockstar&rdquo;, for 2 years in a row.

&nbsp;

<blockquote>
Why did you choose to be a software engineer?
</blockquote>

I have a deep love for code. With the exception of playing with my son, if I had to only write code everyday, I&rsquo;d be a pretty happy man. I studied a degree that was completely different to what I&rsquo;m currently doing mainly because I love the immediate results that comes with coding. Your objectives are attainable, realistic and you basically have the tools to code anything into reality.

When I joined MioTech, results needed to be achieved in a short amount of time. I was given the responsibility of revamping the company&rsquo;s entire framework. It was a daunting, but at the same time a fantastic opportunity. Once you&rsquo;re signed off on the initial stages of conceptualizing, planning and designing the framework, writing the actual code is the easiest part. AMI&rsquo;s framework was achieved in a couple of months and it was truly my proudest moment. My efforts had the desired impact.

<img src="http://miotech-landingpage-images.oss-cn-shanghai.aliyuncs.com/images/1549100387787_Screenshot%202019-02-02%20at%205.39.36%20PM.png/th1000" style="height:auto; max-width:100%; width:auto" />

Starting them young, Yishu brings his son into work.

&nbsp;

<blockquote>
What makes your day not so typical as compared to other software engineers?
</blockquote>

Working in a startup means new surprises everyday. On top of that, you have the &ldquo;Yishu&rdquo;s that come with the nature of software that is errors and bugs. Being able to meet the demand and push out results fast and timely is critical. Right now, we&rsquo;re working on pushing out a new update every week and so far (knock on wood) we&rsquo;ve kept up the pace.

&nbsp;

<blockquote>
What is one myth about software engineers that isn&rsquo;t true?
</blockquote>

Many people think that you need a degree in software engineering in order to be a software engineer. In my experience, I don&rsquo;t think that&rsquo;s necessarily true. It might be a good foundation, it&rsquo;ll get you a foot in the door, but I think 90% of your skills are learnt on the job. I studied something completely different from what I&lsquo;m currently doing so I can say with confidence that what&rsquo;s needed is a love for learning, independence, logic and a critical mind.

<img src="http://miotech-landingpage-images.oss-cn-shanghai.aliyuncs.com/images/1549100466042_Screenshot%202019-02-02%20at%205.39.43%20PM.png/th1000" style="height:auto; max-width:100%; width:auto" />

Yishu&#39;s wife and son at a recent trip

<blockquote>
Why did you decide to work for MioTech?
</blockquote>

Before joining MioTech, I was working for a corporate giant where I felt like a small screw in a larger machine. As long as the task assigned to me was completed and no mistakes were made, the operation of the larger network could be guaranteed. But the longer I stayed, the more I realised the little I could achieve. My work was in part limited to the boundaries set by a fairly entrenched system. It was hard to feel like I belonged.

Initially, while I wasn&rsquo;t particularly clear on what it took to start a business, what motivated me to jump into a startup despite MioTech being fairly small at the time, was the company direction and atmosphere. I had many long discussions with the CTO before I joined and I was inspired by the promise of this business and assured by the talent and reliability of the people at MioTech. I&rsquo;ve only grown more convinced of my choice as time passes.

<blockquote>
What are your key responsibilities?
</blockquote>

I&rsquo;m basically MioTech&rsquo;s front end warrior. Besides the daily grind of putting out software pests, my key responsibilities are to make into reality every business need, fulfill the design fantaries of the product team, and ultimately, code it into a functioning software system.

<blockquote>
What was the biggest challenge you faced at MioTech?
</blockquote>

The biggest change I faced upon joining MioTech was the fast pace. New features that used to take months or even a year to go live in a big company can usually be achieved in a few weeks. Fast-paced product iterations means continuous learning, continuous exploration, continuous trial and error, continuous improvement, continuous acceptance of information explosion which at times conflicts with the idea of speed.

<img src="http://miotech-landingpage-images.oss-cn-shanghai.aliyuncs.com/images/1549100567371_Screenshot%202019-02-02%20at%205.42.40%20PM.png/th1000" style="height:auto; max-width:100%; width:auto" />

Yishu and team debugging at a restaurant in Boracay during a team building trip.

&nbsp;

<blockquote>
What do you value most working at MioTech?
</blockquote>

One thing I value the most is the fact that I&rsquo;m not treated like some code churning machine. I&rsquo;m not writing code for codes sake. I feel that everyone at MioTech feels a sense of ownership for the product. We liken the product and the system to our own child. While we have a collective sense of responsibility, we know that it&rsquo;s our individual performance as an employee that directly impacts the performance of the product. You can see that effort being brought into fruition. What MioTech needs is a person who can pair ambition along with current technological realities. Dreams can also be down to earth as well.

&nbsp;

Hobbies：Coding, Photography, Travel

Yishu&rsquo;s Wish：That his son will also come to love code.

If you could give yourself 2 labels, what would it be？Tech, Perfectionism.

What kind of engineer is MioTech looking for？A passionate, deep understanding individual who also pays attention to details.

Tagline: The code finance needs to be cool.

At a FinTech/AI company like MioTech, (especially since Tech is the second half of our company name), having a good software engineer is as good as gold. Almost 80% of our company is made up of engineers.

My Not So Typical Day as a Software Engineer

Head of Applications

In this article, MioTech invites E. Schoeneburg, an AI Pioneer, AI Visionary and Alternative AI Leader to talk about Deep Learning. He tries to give an objective assessment and judgement about the use, advantages/disadvantages of the various AI models.

Technology

Chinese NLP 101

Business Intelligence 101

The Source of all Knowledge

My Not So Typical Day as a Front-End Engineer

My Not So Typical Day as a Software Engineer

Financial Time-Series Predictions and AI Models (Part 1): Deep Learning - or not?

Technology

Chinese NLP 101

Business Intelligence 101

The Source of all Knowledge

My Not So Typical Day as a Front-End Engineer

My Not So Typical Day as a Software Engineer

Financial Time-Series Predictions and AI Models (Part 1): Deep Learning - or not?

Subscribe to Inbox Insights

Subscribe to Inbox Insights

Subscribe to
Inbox Insights