Keywords

1 Introduction

In recent years, the rapid development of Internet brings geometric growth to web information, and the vast volume of information means it has the characteristic of polynary and redundant as well. Web pages cannot be directly made use of by traditional database systems for its semi-structured characteristic [1]. How to use the information better becomes the focus of attention. At present, most web pages are given to the semi-structured document in HTML form. Web documents can be represented as unstructured documents, semi-structured document and structured document. For the correct extraction of web information, a lot of work has been done at home and abroad. References [2] proposed that results of participle algorithm could touch the shopkeepers’ minds, and it can support the originality data for the commodities markets and dynamic trend analysis. Reference [3] describe the new API for data mining proposed by Microsoft as extensions to OLE DB standard. Reference [4] describe data mining and brainstorms on its application to power systems. Reference [5] supported some actions to promote information and communications technologies in the autonomous community of the region of Murcia of Spain. Reference [6] analyzed the impact of urbanization on regional flood risk in the Qinhuai River Basin of China. Reference [7] applied research of the ash connection analysis in highway’s influence of regional economies. But all of them not reported the useful method to discover new regional investment project using web mining technologies. So our proposed is how to find those new regional investment project and help the modern micro business to solve their concerned at the recent development.

Based on our past work [8–12], we select investor’s discovery and web mining pages from the well-known colleges and universities in China in order to build an efficient system of new regional investor’s discovery by web mining and study the Multi-factor matching method for information ex-traction. The second part gives the system architecture for new regional investor’s discovery. In the third part we introduced web crawler strategy of webpage search. Multi-factor matching algorithm is proposed in the fourth part. In the last two parts we gave application system structure and result of system running.

2 System Design

2.1 System Framework and Development of Language

How to seek for specific investment projects investors from the web, mining data, and store in the database is serious problem for micro businesses. Fast and accurately discovering relevant project information and contacting information is very conducive for micro businesses to publicize and sell their products or services. To build a data mining system, the system should conclude the following functions:

  1. 1.

    Automatically mining the release date, URL and related data information. Such as, project overview, project introduction, project approval document and so on.

  2. 2.

    According to their own requirements and keyword, querying the data in the database easily through software interface.

  3. 3.

    Study of the basic knowledge of Python language.

  4. 4.

    Study of re-module in the regular expression matching.

  5. 5.

    Study of Django web development framework.

  6. 6.

    Study of MySQL data construction and query.

Django’s advantage is simply and rapidly developed database-driven website. It emphasizes the importance of code reuse. Multiple components can be very convenient serving for the whole frame as plug-in form. Django has many powerful the third party plug-in. It makes Django Strong expansibility. It also stressed the rapid development and the principle of DRY.

With Python class form defining data model, ORM related model and database together. You will get a database API using very easily, and also you can use the original SQL statement in the Django language.URL assignment use regular expression matching URL. You can design URL random without specific limited framework.

Using Django powerful and extensible template language, the template system can be separated from design, content and Python code. And it also has inheritance.

2.2 Web Crawler

Web crawler crawls web information automatically following certain rules. Webpage search strategy can be divided into depth-first, breadth-first and best-first. Depth-first in many cases will lead to crawler trapped problems. And breadth-first and best-first are popular methods.

Breadth-first search strategy means that next level search must be after the completion of the current level in the crawl process. The algorithm design and implementation is relatively simple. Algorithm flow is shown as Fig. 92.1.

Fig. 92.1
figure 1figure 1

Breadth-first search algorithm flow

Full size image

Best-first search strategy according to the web analytics algorithms to predict the similarity of the candidate URL and landing page, or correlation with the theme, and select the best rated one or several URLs to crawl. The only access through web analysis algorithm predicted “useful” pages. One problem is that reptiles crawl path on many relevant pages may be ignored, because the best-first strategy is a local optimum search algorithm. Therefore it needs to be combined with the best-first specific application to improve, to jump out of local minima.

2.3 Multi-Factor Matching Algorithm

This paper focuses on the example of chart showing for science expert information extraction.

Let the view-source of webpage be defined as D, body of the page be defined as S:

$$ S \subset D $$
(92.1)

Let the normalized text be defined as \( \hat{S} \):

$$ \hat{S} \subseteq S $$
(92.2)

Keywords corpus consists of some self-learned keyword. Let the keywords corpus be defined as \( F \), then \( F \) be presented as

$$ F = \left\{ {f_{1} ,f_{2} , \ldots ,f_{n} } \right\} $$
(92.3)

In fact, the expert information field included in \( F \). Let it defined as \( f_{n} \), the position of \( f_{n} \) in \( \hat{S} \) is defined as \( k \):

$$ k = (f_{n} ,\hat{S}) $$
(92.4)

Let the potential expert information field be defined as \( t \), define a constant as con then \( t \) is equals to the normalized text range between \( k \) and \( k + con \):

$$ t = \hat{S}(k,k + con) $$
(92.5)

Finally, because of other fields affect, we must remove the affect. The positions of \( F \) except \( f_{n} \) is defined as K:

$$ K = \left\{ {k_{1} ,k_{2} , \ldots ,k_{n} } \right\} $$
(92.6)

Let define the min position in K as \( k_{\hbox{min} } \), define the real expert information as \( t_{real} \), we can conclude that the real expert information filed:

$$ t_{real} = t(0,k_{\hbox{min} } ) $$
(92.7)

2.4 Data Table

System data sheet is mainly used for storage government network project information and credit information publicly shared on all project information in the column. And it could be called and inquired by the system (Table 92.1).

Table 92.1 System data table
Full size table

2.5 System Data Structure

Application system design includes eight sheets, respectively as the auth_message sheet, django_conten_type sheet, auth_user sheet, auth_permission sheet, auth_user_user_permission sheet, auth_group_permission sheet, auth_user_group sheet and auth_group sheet. Its structure is shown in Fig. 92.2.

Fig. 92.2
figure 2figure 2

Data structure of the application system

Full size image

The data structure of application system is succinct than the other system. Our aim is adequate to the boss of micro business for application requirements. If using the web service technology, we can get the encapsulation web service [11, 12] for other application system and get the advantage of platform irrespective.

3 System Implementations

Each hyperlink function in the interface of the application system as follows: URL: Mining the demand URL in the webpage, and store in the URL sheet of the database. Data: Release date of mining project, and store in the date sheet of the database. Info: mining project summary, approval documents, posting date, and store in the Huaian sheet of the database. Query: query the project data and URL in the database. Project Analysis: Building materials can be divided into structural materials, decoration materials and some special materials. Structural materials include wood, bamboo, stone, cement, concrete, metal, brick, ceramics, glass, engineering plastics, composite materials, etc. Decorative materials include a variety of coatings, paints, coatings, veneer, colored tile, glass and other special effects. Special material includes waterproof, moisture-proof, anti-corrosion, fire, flame retardant, sound insulation, heat insulation, thermal insulation, sealing and so on. Figure 92.3 illustrates the connection information in a part of the URL database.

Fig. 92.3
figure 3figure 3

Portion of URL extracted from the database

Full size image

With the using of software, micro business based on building materials can find associated with their investment projects, promote and sell their products and services, thus derive profits to promote their own development.

Take Huaian of Jiangsu province for example, with the investment project comparison of nine counties in Huaian, micro business get an overall Huaian investment regional distribution map. With it, micro business promotes and market their product or service through its own geographical advantages. Figure 92.4 shows the investment project numbers of each region.

Fig. 92.4
figure 4figure 4

Statistical data of nine counties

Full size image

Take the industry for example, culture construction is the development of education, science, literature and art, the press and publishing, radio, television, sports and public health, library, museums and other cultural undertakings of activities. Micro business related with the cultural construction, find suitable for their own projects according to the search results, thus promote and sell their products or services. Figure 92.5 shows the investment project status according to the industry classification.

Fig. 92.5
figure 5figure 5

The investment project status of various industry

Full size image

4 Conclusion

Seeking for specific investment projects investors from the web, mining data, and storing in the database are serious problems for micro businesses. With the technology of data mining, discovering the new regional investment projects targeted the promotion of enterprise products or services can reduce the operation cost of small and micro businesses. The proposed systems can satisfy small and micro businesses for informatization and can meet market requirements.

Python language, MySQL database and Django web framework are used to develop the proposed system which can extract the region’s investment project information from government websites. All the information of investment project can be stored in the database.