In the early morning of September 1st, Wenxin Yiyan was officially opened to the whole society, and users could download the "Wen Xin Yiyan APP" or log in to the "Wen Xin Yiyan official website" on the App Store and Android App Store to experience. Previously, you had to queue up to receive the qualification for the internal test. According to the official introduction, Wenxin Yiyan is a new generation of Baidu's knowledge enhancement big language model, which can interact with people, answer questions, assist in creation, and help people obtain information, knowledge and inspiration efficiently and conveniently.
I would like to know the level of Wen Xin Yiyan in the field of data governance, the same data governance problem, "Wen Xin Yiyan" vs ChatGPT4, who is better of the two? With a point, I can easily infer the level that Wen Xin's words can achieve in other professional fields. Here it is assumed that ChatGPT4 is the benchmark score of 100 points.
Question 1: What does metadata mean?
- Wen Xin's answer:
- ChatGPT4's answer:
It can be seen from the results that the answer language organization of ChatGPT4 is logical, very specific, and more understandable, "Wen Xin Yiyan" feels like copying the definition on the Internet, and then piecing it together, "Wen Xin Yiyan" scored 50 points for the round.
Question 2: How to explain the concept of metadata to a five-year-old?
- Wen Xin's answer:
- ChatGPT4's answer:
From the results, it can be seen that the answers of "Wen Xin Yiyan" and ChatGPT4 are acceptable, ChatGPT4 seems to have evolved, it even added abstract metadata such as how many toys, "Wen Xin Yiyan" scored 80 points for this round.
Question 3: What is the difference between metadata and tags? illustrate
- Wen Xin's answer:
- ChatGPT4's answer:
It can be seen from the results that "Wen Xin Yiyan" compared to ChatGPT4's answer, that is, metadata is an inherent attribute of the data, tags are user-defined elements, used to mark objects, directly point out the essential difference between the two, "Wen Xin Yiyan" scored 120 points in the round, I remember that ChatGPT4 originally answered very well, but this time the performance was not good.
Question 4: What is the difference between metadata and data dictionary? illustrate
- Wen Xin's answer:
- ChatGPT4's answer:
It can be seen from the results that ChatGPT4 clearly points out the difference between the use and scope of application of metadata and data dictionaries, while "Wen Xin Yiyan" just repeats the respective definitions, without making an abstract summary of the difference, and "Wen Xin Yiyan" scored 70 points for this round.
Q5: What is the difference between metadata and data standards? illustrate
- Wen Xin's answer:
- ChatGPT4's answer:
It can be seen from the results that although "Wen Xin Yiyan" roughly knows the definition, it obviously does not understand the concept of standards, because it treats the basic indicators and calculation indicators as normative constraints, and the indicators are only the way they are presented, and whether they are standards actually have nothing to do with it, standards are actually very popular knowledge, "Wen Xin Yiyan" is still lacking in correlation and reasoning ability, and ChatGPT4 obviously understands thoroughly, from the examples it gives to know that "Wen Xin Yiyan" scored 50 points for this round.
Question 6: What is the difference between metadata and metamodel? illustrate
- Wen Xin's answer:
- ChatGPT4's answer:
It can be seen from the results that "Wen Xin Yiyan" only gives a definition, and the gourd is drawn as such. ChatGPT4 obviously has its own understanding, and "Wen Xin Yiyan" scored 50 points for this round.
Question 7: What does metadata have to do with data weaving?
- Wen Xin's answer:
- ChatGPT4's answer:
It can be seen from the results that due to the relatively late emergence of the concept of data weaving, ChatGPT4 could not answer, forced reasoning, "Wen Xin Yiyan" took advantage of time, gave a definition, or OK, "Wen Xin Yiyan" scored 130 points in this round.
Q8: What is the difference between metadata and active metadata? illustrate
- Wen Xin's answer:
- ChatGPT4's answer:
As can be seen from the results, I am not very satisfied, everyone only said the appearance, but did not point out the essential difference, so the difference is not large. Active metadata is a special type of metadata that refers to metadata that is actively collected and recorded during data production. In contrast, passive metadata, that is, metadata that is automatically generated during the use of data, "Wen Xin Yiyan" scored 90 points for this round.
Q9: What is the essential difference between data governance and data management?
- Wen Xin's answer:
- ChatGPT4's answer:
When encountering this kind of problem, "Wen Xin Yi Word" must be ignorant, I have been thinking why not break the word and reason, ChatGPT4 when it says "data governance is "what should be done", and data management is "how to actually do it". Data governance provides frameworks, guidance, and standards for data management"What else do you have to find fault with when it comes to such classic words? "Wen Xin Yiyan" scored 20 points in this round.
Question 10: What does master data mean?
- Wen Xin's answer:
- ChatGPT4's answer:
It can be seen from the results that both give definitions and examples, which can be regarded as a tie, "Wen Xin Yiyan" is relatively simple, and "Wen Xin Yiyan" scored 80 points for the round.
Question 11: What are the characteristics of master data?
- Wen Xin's answer:
- ChatGPT4's answer:
As can be seen from the results, both give the three core characteristics of the master data, criticality, sharing and persistence, and "Wen Xin Yiyan" scored 90 points for this round.
Q12: What is Master Data Management?
- Wen Xin's answer:
- ChatGPT4's answer:
It roughly knows the goal of master data management, but it is not clear what specific measures are taken, ChatGPT4 lists everything you can think of, and exceeds expectations, such as data governance and parameter data management, "Wen Xin Yiyan" scored 60 points for this round.
Q13: The essential difference between master data management system and ODS
- Wen Xin's answer:
- ChatGPT4's answer:
It can be seen from the results that ChatGPT4 has methodological guidance in answering questions, very pyramid structure, good at elaborating from multiple aspects, trying to cover comprehensively, "Wen Xin Yiyan" answers the question a little unorganized, think of wherever you say, it does not fully understand the purpose of ODS, "Wen Xin Yiyan" The round scored 50 points.
Q14: What is a data object? illustrate
- Wen Xin's answer:
- ChatGPT4's answer:
It can be seen from the results that the case given by ChatGPT4 is very detailed and easy to understand, although "Wen Xin Yiyan" also gives a definition, but the example given is wrong, "Wen Xin Yiyan" scored 30 points for the round.
Q15: What is the difference between an object and metadata? illustrate
- Wen Xin's answer:
- ChatGPT4's answer:
It can only be said that this question exceeds the ability of "Wen Xin Yiyan", and "Wen Xin Yiyan" scored 10 points in this round.
Question 16: What is a data schema, with examples
- Wen Xin's answer:
- ChatGPT4's answer:
Data architecture is not actually a very unfamiliar concept, but it completely surpasses the answer ability of "Wen Xin Yiyan", and "Wen Xin Yiyan" gets 0 points in this round.
Q17: When is it appropriate for enterprises to start data governance?
- Wen Xin's answer:
- ChatGPT4's answer:
It can only be said that this problem surpasses the ability of "Wen Xin Yiyan", not only suddenly jumps out of the inexplicable terms S2, S4, but also the text is not well organized, "Wen Xin Yiyan" scored 0 points in this round.
Q18: Talk about the essential differences between data warehouse, big data platform, data lake, and data middle office, try to be as concise as possible, and give examples of the best
- Wen Xin's answer:
- ChatGPT4's answer:
From the results, it can be seen that "Wen Xin Yiyan" is not actually talking about the difference, but listing the concepts, and some contradictions, although ChatGPT4 is also talking about concepts, but pay attention to the perspective of each concept is consistent, you can experience the nuances between each other from the explanation of these concepts, and finally there is a summary, I like it, data warehouse focuses on centralized processing, big data platform focuses on diversified data processing, data lake focuses on centralized storage, data middle office focuses on data services. "Wen Xin Yiyan" scored 30 points in this round.
Question 19: Please identify the connections and differences between data, data resources, data assets and data elements
- Wen Xin's answer:
- ChatGPT4's answer:
From the answer of ChatGPT4, it is easy to distinguish the connection and difference between the four, data elements constitute data resources, important data resources constitute data assets, and all available data constitutes data resources. Data is the foundation of this system, data elements are components, and data resources and data assets are collections of different granularities. "Wen Xin Yiyan" did not explain the concept of data elements clearly, and even confused the difference between data elements and data assets, and "Wen Xin Yiyan" scored 30 points for this round.
Q20: What is the essential difference between data indicators and data labels?
- Wen Xin's answer:
- ChatGPT4's answer:
As can be seen from the results, both point out the essence, ChatGPT4 is more detailed and detailed, and "Wen Xin Yiyan" scored 80 points in this round.
After 20 questions were asked, compared to ChatGPT4, the average score of "Wen Xin Yiyan" was 56 points. At least 90% of the answers to ChatGPT4 can be summed up by richness, 30% is of some value to me, 10% is incremental, and 5% will be a stroke. 10% + 5% is exactly 15%, which is the golden ratio of learning.
From point to point, we can make an inference that in the professional field, there is still a big gap between "Wen Xin Yiyan" and ChatGPT4; Of course, in literature, art or history, etc., "Wen Xin Yiyan" is estimated to perform much better, which has a lot to do with Chinese corpus.
In the professional world, there is much more corpus in the English-speaking world than in the Chinese, and ChatGPT4 will convert Chinese to English when answering Chinese questions, resulting in higher quality results. Think about it, in addition to Zhihu and CSDN, there are several websites that can obtain professional knowledge for free?
In any case, "Wen Xin Yiyan" still needs to be cheered!