Penn Treebank Format Parsing Grammar

start with tree_list

tree_list := tree
           | tree_list tree
tree      := ( node )

node      := leaf
           | TOKEN node_list
leaf      := TOKEN TOKEN
           | leaf TOKEN
node_list := ( node )

           | node_list ( node )
TOKEN     := ([^\)\(\r\n \t]+)


Dowload: JLex code, JCup code

Compile SRILM

1. mkdir %SRILM%
2. mv srilm.tgz %SRILM%
3. cd %SRILM%
4. chmod -R 777 .
5. modify Makefile
    5.1 $ sbin/machine-type, suppose it returns i686
    5.2 in common/Makefile.machine.i686
          GCC_FLAGS = -m64 -mtune=i686 -Wreturn-type -Wimplicit
          insert NO_TCL = X
          delete TCL_INCLUDE, TCL_LIBRARY
6. make MACHINE_TYPE=i686-m64 World

verb list whose VBN and VBD are same

matched
developed
compared
reduced
created
correlated
detected
demonstrated
genotyped
indicated
observed
fed
disrupted
repressed
transduced
designed
examined
investigated
prevented
suggested
produced
revealed
identified
affected
consisted
presented
assessed
abolished
tested
injected
resulted
determined
resected
confirmed
reported
measured
performed
varied
supported
aimed
collected
evaluated
allowed
inhibited
used
caused
analyzed
amplified
decreased
expressed
studied
transfected
included
increased
contained
encoded
induced
applied
found

Noun Phrase Similarity Rules

Observation:

1. integer: [-+]?\d+
2. rational number: \d+[/]\d+
3. finite decimal representation: [-+]?[0-9]*[.][0-9]+([eE][-+]?[0-9]+)?
4. suffix hyphen: \w+[-]
5. prefix hyphen: [-]\w+
6. infix hyphen: \w+[-]\w+
7. [A-Z]{2,}
8. [A-Z]+\d+
9. [a-z]+\d+
10. [a-z]+[A-Z]+
11. [a-z]+[A-Z]+\d+
12. [A-Z]+[a-z]+\d+
13. [a-z]+\d+[A-Z]+
14. [a-z]+\d+[a-z]+\d+
15. [A-Z]+[a-z]+[A-Z]+\d+

Summary:

1. integer: [-+]?\d+
2. rational number: \d+[/]\d+
3. finite decimal representation: [-+]?[0-9]*[.][0-9]+([eE][-+]?[0-9]+)?
4. suffix hyphen: \w+[-]
5. prefix hyphen: [-]\w+
6. infix hyphen: \w+[-]\w+
7. [A-Z]{2,}
8. [A-Z][\w \\\/-]*\d
9. [a-z][\w \\\/-]*\d
10. [a-z0-9]+[A-Z]+
11. [a-zA-Z]+\d+[a-zA-Z]+

counter example of combining 3 and 4:  
p53 and TAFII40 and TAFII60

Varying Sentence Beginnings

1. Non-finite and verbless clause
1.1. gerund-participle
Using histidine - tagged recombinant TR6 , we screened soluble forms of TNF - ligand proteins with immunoprecipitation .

counter example:
Immunoblotting experiments indicate that GRB2 associates with ...

1.2. past-participle
Based on the experiences of all the authors , we emphasized cell cycle inhibitors such as p16 and p21 and regulators of apoptosis such as p53 and members of the bcl - 2 family .

1.3. to-infinitival
To examine whether GRB2 is also a component of ras signaling in mammalian cells , microinjection studies were performed .

2. dependent clause
As previously shown with reticulocyte lysate - reconstituted steroid receptor heteroprotein complexes , the reconstituted pp60 src multiprotein complex contains hsp70

Although both unstimulated and activated human T cells express high affinity IL - 7R , only activated T cells can proliferate to IL - 7 .

3. preposition phrase
In addition , hsp90 in the lysate complexes with wild - type pp60v - src .

From Scatchard analyses , the Kd :s for binding of aFGF and bFGF to hFGFR - 1 were estimated to 25 pM and 41 pM , respectively .

4. adverb
Interestingly , GRB2 exhibits striking structural and functional homology to the C. elegans protein sem - 5 .

5. noun phrase (an object)
The letter , I read it to to you.

Reference

1. Rodney Huddleston, Geoffrey Pullum, etc. The Cambridge Grammar of the English Language. page 51,  604, 1173
2. http://intensive-english.blogspot.com/2009/01/25-varying-sentence-beginnings.html