The former cannot fear my own achievements (on the inspecialization of product search engines that have already been built) [original]

// Author: chengg0769 blog.csdn.net/chengg0769 (ensure text integrity during reprinting)

Tonight seriously studied the business search: http://www.shangsou.com evaluation of the effect.

Why study it, because it "is still okay", other products or commercial search, the traffic is less than 1/10 of it, almost in addition to the impetuous voice, is not worth seeing.

1. for product query, take "Mouse wire" as an example:

Supply tab result: 1

Results on the purchase tab: 11

Seller tab results: 5

Product tab result: 1

Image tab result: 0

So few results are surprising, at least I am a person who has been in the factory for a few years. Haha

2. Fuzzy search with "bright" as the company name

Results On the merchants tab: 9958, no region or industry category. If I want to find a company in Dongguan, I need to finish reading 9958. Enterprises in different places cannot be seen at a glance, and the word may be included in a situation similar to "Sunny.

Of course, the company name itself is not a commonly used word, and there is no way to achieve this.

3. Extraction of company profile and contact information

Based on my own company, the results are not extracted from the official webpage of our company, but from the information published by a salesman on a small website, the salesman has been away for three years. The number is 7 bits, and now we have changed it to eight bits for half a year. In theory, enterprise websites must first be captured when establishing enterprise databases. For my analysis, the basic information provided by enterprises is often very different, it is certainly not easy to copy the information of some directory-based websites that comply with certain templates. Company virtual shop number used in shangsou.com with prefix domain name, such as Chongqing shanlun Motorcycle sales Company Limited: http://cnslmtc.cn.shangsou.com/conjecture, so nonstandard numbering method, can only be manually completed, huh, huh. Some are pinyin, some are English, some are Pinyin and English, and some are Pinyin and serial numbers. If it is assumed that it is artificial assistance, you still need some perseverance!

4. Supply and Demand Information

The results depend too much on the results of capturing Alibaba, while the information published on other websites is less captured. (I have not verified much)

5. Wishful thinking of shop settings

It can be seen that stores only generate a second-level domain name based on the enterprise name, and place the original course materials there. I have verified several companies, as if they did not participate in the interaction between enterprises, to put it bluntly, no one takes care of it. Errors or errors. Many enterprises have websites. I don't know whether they are technical problems or other problems. Even the most basic official websites of enterprises cannot be provided completely.

6. Information integrity

Users may not only care about the most basic contact information, but also a brief introduction. If users are buying products, they are more concerned about Enterprise Products and Services, enterprise development and design capabilities, and manufacturing capabilities. Concerned about its certification and environmental protection.

7. closed user participation Mode

If users do not interact with each other, there are almost no sources of Self-information except crawling, because it is not easy to involve enterprises in your own libraries, Because you implement membership, we must talk about money. In addition, this method is more mixed with the B2B model trend.

8. Keep the source link

No matter where the information comes from, it is the most fair attitude to allow users to directly access the original link under any circumstances. By extracting so many enterprise data and blocking source links, this spirit is contrary to search, but instead it is necessary to search for data. I think this is not true for copyright and fairness.

9. service loss

In the early stages of my business, it would be necessary to attract users as much as possible and serve enterprises in the database. At the beginning, I would seek the greatest benefit, which should be undesirable and will not work.

========================================================== ======================================

To sum up

1. vertical search I don't think this is the way to do it. I don't know the user's needs. I want to do it with a concept. For users, soft texts have no effect. They only want results. Otherwise, they just despise the results. Like my current attitude.

2. vertical search, enterprise database, and product search. If you have no experience or knowledge, how can you implement the concept. There is no concept of products, no concept of manufacturing, no industry experience, no professional, semi-professional participation. How can this problem be solved ??

3. There is no demand in this regard. From the current Alexa analysis, we can see that the traffic is not small. My estimate is at least 60-80 WIP/day. So if you want to enter, there is still space. However, if it is not the result of a data capture task, but the task is captured every day and updated every day, you need at least 30-50 computers to capture and analyze the task. It is impossible to make a profit very quickly. Who knows how much money can be supported by their own gold coins? Of course, we have to work with good algorithms and content processing methods. Otherwise, it is unavoidable to defeat the ground.

4. In some aspects, this seems to be the content. Of course, the technology is also critical and must be continuously improved.

Comments are for reference only.