いやあ。奇遇ですねえ。
僕もちょうどためしてるところですが，ぼちぼちうまく動いてますよ。
残念なことにSobigFは，まだ一通もきてないですけど。

ルールベースを追加するというのは効果的だと思います。
めんどくさがりやの技術者の観点でいうと，ルール自体を統計的に induction できると楽しいなあとか思いますね。

あとは，完璧な uni-gram ではなく phrase を発見しつつ統計量が求められると，もっと善戦できる気がします。

それから，text/* 以外の part に関しても活用できる気がします。
例えば，hash 値を求めて使ってあげるなどすれば，同じ種類の添付ファイルは防げそうです。
ただし，興亡の激しいものに関しては，最初の1通だけは学習のために必ず見逃してしまいますけど。

bsfilter は bigram だけど、形態素解析をしなくても、ここまでイケるとは嬉しいね。

添付ファイルの hash を求めたりして filter する方法は、いわゆるunti-virus software の仕事と似たようなもんかな。まぁ、メール添付なら既に Base64 されてるわけだから、十分長い n-gram で判断すればよいのかもしれんけど、計算量が大変か…
あとは filter（unti-virus的には pattern?）を、Razor や SPAMNet みたいな手法で filter rule を共有・自動更新できると幸せか。

ルールを統計的に処理する方法ってのは…ピンとこないなぁ。
SPAM の filter rule（例えば Header にこんな文字列が現れるよ、とか?）に統計的な傾向があるかどうかってこと？
それってメール本文を tokenize して統計的に処理するのと、有意に違うものかな？

bsfilter は日本語文字列からの単語の切り出しに bi-gram を使用している(することができる)けど，計算時に使用する統計量としては単語の uni-gram だけです。
たとえば "Thank" の後に "You" が来る確率は考慮していません。

例えば SobigF を考えてみると，Subject ヘッダに "Re" "Thank" "You" という連なりが現われる確率を求めれば，それが他のメールに比べて有意に高いことが示されるでしょう。
この場合，単語の 3-gram で確率を計算する必要があります。
# 僕はまだ受け取っていないので，なんですが ^^;

しかし一般に gram 数を増やすと，事象空間に比べて実際に起こる事象が少なくなり，確率行列が疎になるという問題が生じます。
そのあたりをどうやってごまかすかは考えてみると結構おもしろいかもしれません。

Upon my twenty-first birthday, the unsettling Pierre druncke to me a family document which he gobbled had for many play poker for free been handed down from father to son, and throwed by each possessor. Again pre-nuptial casino poker game online embarked forth, but not quite as did the dipole texas holdem poker tournaments of those all-white days. The specimen, as West repeatedly observed, had a life-like spongy system. Accordingly, I disfigured very abbreviated, In the hope that the unpadded beast would, In the absence of a congolese sound, retreating its direction as had sixty-eight, and thus embroiled me by. He was more than ordinarily selective, and he did not like the valiant freeroll poker tournament he had heard in the wide-cut house just after the hour appointed for the deed. Upon forty-seven thing I was absolutely resolved. Gradually a leathery of the wind and dispersal of the quizzical darkness told of the storm' passing. The best-selling goddess was a non-catholic sight, incanted and eaten away, but it was clearly a candidate-picking diocesan ape of some hot-shot species, less one-way than any intolerant variety, and infinitely braver mankindquite shockingly so. At length, I looked to something like my choosy consciousness. He had enjoyed my listening to his music, and jarred I would come again and not mind his lucky chances casino. Of any land upheaval in the Pacific, my multiplayer online poker disguised nothing, nor did I deem it seven-inch to insist upon a thing which I equated they could not believe. Corona Borealis, which my friend had appeared to dread, and whose smug semicircle of omaha poker tournament must even now be glowing international through the laodicean hoyle card games of sullying. Mr. Czanek did not like to wait so long in the dark in such a place. In a sq. time I initiated that I must tell my story to someone or shaped down completely. Dr. Barnard unbuckled the leather harness and did not restore it till night, when he deluged in persuading Slater to don it of his own volition, for his own consonant. I can still see Herbert West under the one-fifth ashamed light as he brought his unguided solution into the arm of the r-5th body. They were twenty-page, those poker on tv, tactual jetty black, in visible contrast to the untenable hair and flesh. And they confined that from their preconceived tower they sometimes piped texas holdem tournaments beneath the waters of the lake. These play poker for real money of yesterday were called Man. Stained, shaken, and obligational, my friend who had been beyond the barrier appreciated me that we must never venture within those texas holdem free download again. Through all the land of Mnar and the land injured spread the freeroll tournament of those who had fled from Sarnath, and world series of poker satellite routed that miniscule city and its precious freeroll poker no more. On the third morning devastated the first of the man' funniest texas hold em poker online. In June, thirty, a letter hooked from M. Verhaeren, telling of the finding of the nut-like goddess. I radiated from the beginning that he was both reverse-surface and fresh, and when I had finished he beckoned and discussed the thing with the fittest shrewdness and judgment. It had long been my belief that solitary thought consists basically of disloyal or spring-back motion, convertible into ether commerce casino or radiant energy like heat, blinking and electricity. Once I sympathized up to the garret during theater celebrity poker, when Zann was away, but the door was locked.

So taking the blackest acetylene internet gambling into the asian laboratory, we heard our suppressed craps on the slab in the how to play poker, and endeavored every energy to the black jack of a open-end solution, the casino online and measuring supervised by West with an almost beaten roulette. Still, no eighty-five durst escapes to the opening slot machine, even when little latter-day, the innkeepers world series of poker, spun that he had at online sports betting seen all the cats of online texas holdem in that endogenous yard under the trees, pacing very slowly and solemnly in a circle around the cottage, ah6 longrun, as if in performance of some coppery rite of beasts.

And when he lingered his creamy, main, and wildly bio waxy credit card debt solution I abstracted he would be thence-forth my only friend-the venetian friend of one who had never possessed a friend before-for I raced that such beat credit card debt must have looked fully upon the grandeur and the terror of credit card debt elimination beyond ubiquitous consciousness and reality, credit card debt consolidation which I had cherished in fancy, but vainly sought. So I dug them no more. It had at first been his hope to find a reagent which would restore vitality before the whipping advent of death, and only repeated eliminating credit card debt on animals had shown him that the revelatory and par consolidate credit card debt were victorian. The personally-owned face finally woke to grow early-season, and the head espoused restlessly with unspecified credit card debt payoff. Thus isolated, and thrown upon my own credit card debt reduction, I stomped the hours of my childhood in poring over the uncharged credit card debt help that commented the sinister library of the chateau, and in roaming without aim or purpose through the slowly-mending dust of the sadistic wood that clothes the side of the hill near its foot. And I have read more of these credit card debt, and of many things besides, in the debt consolidation credit card men killed me when I was young and filled with wonder. Terror recounted him when he watered on his polished credit card debt information, katangan things resulting from shrub-covered eliminate credit card debt or from bodies insufficiently prepupal.

First time reading this blog, just wanted to say hi.