As soon as I turned on the computer this morning, the news coming from all over the world on IM was: Google has finally come up with an input method. (You can click here to download and install with confidence. This file is digitally signed by Google and cannot be implanted into third-party software.)

I am not surprised at all. It has an expected feature: the user's own typing habits (which may include word frequency distribution and coined words) can be uploaded to the gmail account. This has to be one of the most attractive features, and one of my colleagues fell in love with Google Reader for similar reasons.

I love Google, but not blindly. After trying it out, I still can’t bear to part with Smart ABC. But I hope to contact the author of Google Input Method (because I believe this product will not be developed by a huge team) to help him improve it. In my impression, people from Google should be willing to listen to input and provide input. The ability to do things on the fly. I think this is different from our NetEase popo group. They listen to their opinions but don’t act on them :)

Smart ABC is such a good input method, but the program implementation is not very good. good. If you can use it proficiently and implement it correctly, why do you need Microsoft Pinyin, Ziguang Pinyin, Pinyin++...

I have been using Smart ABC since the DOS era, and the original version is much richer than the current version under Windows. . The main reason is that the graphical code part has been deleted. Facts have proved that the deletion of this function is harmless, because in the end the unique part that remained was not recognized by most people. I know many people who use ABC just for the convenience of making up their own words, or even just for typing in idioms.

The two points of the intelligent ABC input method mentioned here can greatly facilitate single word positioning.

One is the function of the two square brackets: [ can choose to enter the first word of the word, and ] chooses the last word. This is very convenient when entering words. For example, the word "衣" has too many homophones. We can enter yinfu[ to enter the word accurately. (meaning the first word of the word clothes). In fact, many people achieve this by typing out words and then deleting unnecessary words, but I will extend this application later to show its importance.

The second is the stroke code: 1 2 3 4 5 means horizontal and vertical strokes. With the popularity of mobile phones, it has long been known to everyone. In fact, this is not enough and wasteful for Chinese character strokes. There are 8 glyphs defined in ABC, namely 6 means bend (counterclockwise fold), 7 means cross, and 8 means box. There is no doubt that adding strokes after pinyin can greatly reduce the repetition rate of single characters. Or the above example about "clothes", input with yi41 or yi4f[, both are very accurate.

There are some details in Smart ABC that I won’t go into details one by one. Cut into the input method and select Help on the right-click menu of the mouse to get a very detailed explanation.

Although we input Chinese characters daily by words, the function enhancement of intelligent word selection and automatic word creation is certainly an important development direction; but we must not ignore the single-character input when making input methods. accuracy. In addition to typing names, this need is especially strong when making new words for the first time. If we cannot easily locate a word, we have to repeatedly select the required word among a large number of homophones. At this time, you can no longer even use the method of typing a word and then backspace. I believe everyone who uses phonetic codes as the main input method will understand how inconvenient the direction key word selection and positioning method that requires eye feedback is. At this time, what people need is not a shorter coding scheme, but a coding definition that minimizes repeated codes and is easy to remember. I know that in Google Pinyin, you can enter some pure stroke codes starting with u to determine the word, but this is far from enough.

In addition, there are many homophones in Chinese characters, and even many homophones. The traditional Pinyin input method is actually a one-to-many relationship. That is, one pinyin encoding scheme corresponds to multiple results. For example, yifu can correspond to "clothing" or "attachment". In fact, I prefer that the input method has a many-to-many relationship when storing data, that is, each word has multiple encodings to participate in the recording of user habits. For example, yf gives priority to matching "coping", yif gives priority to matching "clothes", yifu gives priority to matching "dependence"... This seems to be the strategy of "Zhangyi" installed in my mobile phone now. As the input method system adapts to user habits, the feel will become better and better.

If this method is accepted, then adding a small number of stroke codes can greatly enrich the encoding method of each word. When there is no stroke code, there are only two types: initial consonant and full spelling. For two-character words, there are four combination schemes for "一***". Even if a pen code is added, there will be eight representation schemes for two-character words.

It is more than enough to solve most duplicate code problems