Intelligent semantic aggregation framework: Collect and understand knowledge like humans

Zheng Yi 20071129

Intelligent semantic aggregation frameworks are not applicable in any field. They can be applied to the following vertical fields:

• Information Source: The network information is rich enough, with many fragments and scattered;
• User requirements: the audience is increasingly dependent on network fragmentation to form their own opinions;
• Business Needs: easy to establish business models or e-commerce fields.

That is to say, many Internet word-of-mouth companies are either commenting, moving comments, or coming up with some commercial reports or marketing tricks on forums, but they are far from really understanding what people are talking about. In fact, there are a lot of interesting things in semantic applications. It is not just like playing ju.com to make automatic discovery of hot spots like techmeme. After all, the hot stories of playings cannot really understand the intention of the story; for example, the word-of-mouth rankings of Daqi network only involve the computation of the sentiment disposition of text, but it does not understand why a Post says that the product is poor, where it is not good, and why it is necessary to buy the product.

Does the current search understand what people are talking about?

I don't know either.

They may know which keywords you mentioned and which words are more important, but they will not see the knowledge of webpage content like humans. The existing search can only be called "together.

The intelligent semantic aggregation framework is acceptable. What we are currently doing is to try to fully reflect the superiority of this framework in the securities market and tap into the value that only talents can do. At this level, cool News or aibang's life search can only be regarded as "precise information extraction" and "smart organization.

So what is knowledge collection and understanding like humans?

Here is an example of Yahoo search.

This interpersonal relationship map of Yahoo was also copied by Google in the End of the World: http://www.yahoo.cn/s? V = person & P = % E6 % 96% B9 % E8 % 88% 9f % E5 % ad % 90, Fang zhouzi's interpersonal network relationship diagram. We took some time to get started in the last few months. First, we collected news from all kinds of stars, and then trained machines to understand the possible relationships between humans, next, we will count the potential relationships between entities and entities based on the distance between the entity name (which you can understand as the person's name) and the object name and the relationship words, finally, reverse and straighten out the relationship and sum up. The difficulty lies in that the relationship between entities and entities is not necessarily expressed simply by the simple sentence "xxx". Chinese is ever-changing and news is often an event, you need to train enough machines to understand enough relationships.

The understanding of this model is like thinking of people (or children. This is just a simple example.

Summary:

Intelligent semantic aggregation framework. Step 1: select the vertical direction. Step 2: together the knowledge fragments, including blog, news, forum, microblog, and live room; step 3: Text Mining and statistics; Step 4: show value.

Generally, the core issues of computational linguistics and Natural Language Information Processing Research are language understanding and language generation ). The intelligent semantic aggregation framework also belongs to the former secular application. The former identifies the syntactic structure of a sentence from the word and symbol string on the surface of the sentence, judges the semantic relationship between the components, and finally clarifies the meaning of the sentence expression. It has been a long time in the academic world, but to hide the complex technologies behind it and turn it into a general application demand, we still need to come from our real life, observing human daily behavior may be a good way. I am worried that only geek has the need to compete with countless people.

My otherArticle:

11/27/2007 topic marketing can only be the production of garbage and garbage traffic at the moment of freedom of choice
11/23/2007 "Search + Community"
11/01/2007 [lecture] The "superstitious" of Internet people"
11/06/2007 "hit", "hit", and "hit"

The GPS navigation service should be farther away at 11/06/2007.

10/29/2007 [Help] mobile internet's "Muddy Waters"

10/22/2007 SmallIRobots, Msnnext, msnshellWeekend party [help Club]

10/16/2007 Web2.0Information Organization needs to introduce new ideas of Semantics