Slashdot Mirror
WordLogic Patented the Predictive Interface
Packetl055 writes "Have any of you heard anything about this company, WordLogic, with a soon to be granted/issued patent with 117 claims for predictability software? They recently received a patent approval/allowance letter from the US Patent and Trademark Office. Their patent application was submitted in March 2000. If I read this correctly, any software that gives you any prediction after you type something is infringing on their patent — e.g. vehicle navigation systems, cellular telephones, PDA's, Google with their 'Did You Mean' when using Google for a search, the new Apple I-Phone, Blackberry, Sony Playstation-3, etc., etc. If true, this is going to be huge: lawsuits after lawsuits." Their stock trend over the last few days suggests that somebody was paying attention to the the USPTO news from August 9. WordLogic makes products (assistive input software) and doesn't seem to be merely a patent troll.
Prior art from 1996, anyone. Thank you, Bill! ;-)
I live ze unknown. I love ze unknown. I am ze unknown.
Don't assume that there's prior art just because the Slashdot summary seems to be similar to things you used in the past. The only measure of valid prior art (other than actually going to court) is when a patent lawyer looks over both the letter of the claims and the claim of prior art. Often, in that light, the prior art turns out to have no relevance.
Patent submitters typically know about the most obvious examples of prior art, so most patents are worded to carefully carve out a niche in which the patent almost, but not quite, describes existing technologies.
J. Carlberger published a paper titled Design and Implementation of a Probabilistic Word Prediciton Program in 1997. That should pretty much take care of the prior art argument.
http://www.peertopatent.org/ is where it is at, and they have a mailout list that will keep you informed on new patent apps.
Although I haven't seen one where my knowledge could affect the process, the very first time I hear of a patent application that I can attack with prior art, I will do so immediately.
That way things like this patent don't get so damn close to being approved before we can jump on it.
...Open Source isn't the only answer -- but it's almost always a better value than the alternatives...
And even if they haven't been a patent troll in the past, they may be becoming one.
Let's look at their latest quarterly report (SEC filing) and pick a few bits:
So it is a company that is making losses and focuses mainly on a single product. The success of this product depends on the licensing deals related to that patent. It looks like that company is betting a large part of its future on that single patent. So their best hope may be to become a patent troll. It may be a bit sad for the engineers working at that company, but I have serious doubts about their future business plans and methods.
Is it surprising that they issued a press release related to that patent a few days before issuing their quarterly report?
-Raphaël
First patent is using dictionaries in predicting incoming text
1) SYSTEM AND DEVICE FOR PREDICTION OF SUBJECT ( JP1029972 )
PURPOSE:To analyze the content of a text based on prediction, by analyzing an inputted text by using the grammatical rule of a targeted language, predicting the subject of the text from a word possible to regulate the subject, and predicting the subject predicted from the largest number of words as the subject of the inputted text. CONSTITUTION:The titled device is provided with a subject dictionary 1 in which the candidate of the subject predicted from each word is registered in every word unit, and a subject indicating word segmentation part 2 which analyzes the inputted text grammatically and extracts the word to become the main constituent of the input text. At a subject selection part 3, the subject dictionary 1 is referred, and when no subject candidate to be predicted exists in every word unit extracted at the subject indicating word segmentation part 2, no operation is performed, and when it exists, it is taken out, and the number of taking out is held at every taken out subject candidate, and when the taking out and the counting of the number of appearance are completed, the subject candidate having the largest number of appearance out of taken out subject candidates is outputted as the subject of the inputted text. In such a way, the subject of a supplied text can be predicted. The second patent uses previous text inputs in helping to predict the incoming text
2) METHOD AND DEVICE FOR PREDICTING SUBJECT> ( JP2280272 )
PURPOSE:To analyze the content of a text based on prediction by holding the set of micro features having the number of times of appearance exceeding a critical value as the present status, and assuming a subject expressed in the partial set of the micro features most neighboring to the above set as the present subject. CONSTITUTION:A recent appearance word meaning storage means 2 stores meaning by the expression of the micro feature of a constant number of words appearing recently, and a critical value filter 3 delivers only the micro feature in which the number of times of appearance of the micro feature existing in the recent appearance word meaning storage means 2 exceeds the critical value to a present status storage means 4. A most neighoring subject selection means 6 compares the set of the micro features held by the present status storage means 4 with the expression of the micro feature corresponding to individual subject in a subject dictionary 5, and outputs the subject having the least common part as the present subject. In such a manner, the content of the text can be analyzed based on the prediction.
I predict WordLogic's patent application is not viable.
Does anyone here ever read these patents before posting how stupid they supposedly are? Reading the patent clearly shows that they are not applying for a patent on predicting text alone, but rather on an input method used by most PDA's and visual keyboards. In addition to displaying possible words, it HIGHLIGHTS the keys on the visual keyboard or displays the input strokes required to generate the next character that it thinks you want to select. Visual Studio never lit up characters on my keyboard, it never even displayed a visual keyboard on my screen. I'm all for patent reform and striking down obvious techniques, but in this case, while IANAL, on its face value it looks pretty legitmate to me.
(Not entirely sure that its all that much more useful than the standard predictive text stuff that I've already seen or used, but that is not the point.)