xword

[yaoz...@gmail.com]

introducing xword, a new paradigm for online search...

an xword creator can create an xword such as: _NLP := "natural language
processing" OR "computational linguistics"

then, a user can add this xword definition to his xword management
software. after that, he will be able to search: _NLP "word sense
disambiguation",

which actually will search: ("natural language processing" OR
"computational linguistics") "word sense disambiguation"

an xword is like a C macro that encapsulates a complex search criteria
for reuse.

i also devised some more xword features. for example, an xword can
have "primary version", "secondary version", etc. if the primary
version of an xword causes the whole search expression to return too
few search results, it can automatically "relax" itself to the
secondary version and search again.

i will call this feature 'relaxation variable' although it has an
entirely different meaning from the math concept.

an xword is prompted to the user if the user has typed any significant
keyword existing in this xword's definition.

when an xword is used, the computer can also prompt the user of
available related keywords or xwords, which are also defined in the
xword's definition.

Yao Ziyuan

[yaoz...@gmail.com]

natural language (or other liberal arts) shit can't be cured, but can
be improved.

Yao Ziyuan

[yaoz...@gmail.com]

We badly need a search operator "SYN" which indicates a probably direct
syntactic relation between its two operands. This is currently
implemented by expressions like "x * y", "x * * y" and etc.

an xword NEAR_BETTER(x, y) can be defined to do this by providing
multiple versions, with primary version being ("x y") OR ("y x"),
secondary version being ("x * y") OR ("y * x"), etc.

Yao Ziyuan

[yaoz...@gmail.com]

The said xword should be better named NEARER(x, y).

Yao Ziyuan

[yaoz...@gmail.com]

Given what is provided by today's search engines, xword can maximally
automate many re-search attempts. (Although you have to automate it
slowly as not to be recognized as a robot)

Yao Ziyuan

[yaoz...@gmail.com]

We can also have a modifier to indicate that some xwords should relax
faster than other xwords.

Yao Ziyuan

[yaoz...@gmail.com]

xwords can be automatically derived from a person's search history.

Yao Ziyuan

[yaoz...@gmail.com]

The user should be able to select a context in his xword management
software so that only the xwords related to a specific domain are
prompted.

Yao Ziyuan

[yaoz...@gmail.com]

An xword may also automatically narrowing itself to the primary version
if current search results are too many.

Yao Ziyuan

[yaoz...@gmail.com]

* the more primary version.

Yao Ziyuan

[yaoz...@gmail.com]

One important kind of "related xwords" that an xword can lead to is
"subcategory". This way, keyword search becomes table-of-contents
lookup.

Yao Ziyuan

[yaoz...@gmail.com]

An xword as a subcategory:
(1) lets users agree upon a "standardized" term with which they refer
to a subfield in their new documents;
(2) tries to catch the subcategory of documents using carefully
designed search criteria encapsulated into itself.

Yao Ziyuan

[yaoz...@gmail.com]

An xword could also prompt its supercategory xwords.

Yao Ziyuan

[yaoz...@gmail.com]

Is xword a collective effort to standardize a terminology and a
classification? Like Wikipedia is a collective effort to standardize
people's view on knowledge.

Yao Ziyuan

[yaoz...@gmail.com]

If the user wants to find a concept product for which there has not
been a unique category xword, then:

EITHER
The author of that concept product is responsible for making a new
category xword (analogous to a new technical term) for this concept
product and linking this new xword to its supercategory xwords and
other related xwords.

OR
The searcher assumes that any new concept product is the cross
product (mixture) of two or more existing concepts (xwords). If the new
concept product is expressed as many subconcepts and their relations,
it becomes harder to capture the whole concept product accurately.
Instead, we should better view the concept product as a cross product
of two existing concepts. The relation between these two concepts is so
elusive that its discovery is exactly why the innovation of the new
concept product is great. We focus on the expression of the two
existing concepts by xwords and expect to find documents that contain
both of the two existing concepts. Further, we could establish a new
xword as a subcategory xword registered under either of the two
existing concepts (xwords).

Yao Ziyuan

[yaoz...@gmail.com]

without xwords we're unable to search for a complex concept structure
(product) just because we can only describe its constituents at either
a very low level or "a high level without variant forms".

it's like to search for the architecture of a building by describing
its bricks.

Yao Ziyuan

[yaoz...@gmail.com]

"like to search" -> "like searching".

[yaoz...@gmail.com]

Typical search strategies for matching an anticipated concept product
are:

(1) PROBLEM + NEW HOTSPOT:
perspective/context/background/purpose/use/"from what problem could you
come here" xword + another constituent xword that makes the
breakthrough;

(2) WORKING COMPONENTS: xwords representing components that together
make the new mechanism work. Choosing such components (especially
events/conditions and verbal senses) can be tricky because it easily
involves external theorems.

Yao Ziyuan

[yaoz...@gmail.com]

People always introduce / search for something using as few as
possible, as advanced as possible constituent concepts.

Yao Ziyuan

[yaoz...@gmail.com]

If working components alone generate too many results, pairwise
concept-to-concept relations should come in.

Yao Ziyuan

[yaoz...@gmail.com]

The designing process of an xword is not a solely human work. The
computer should make suggestions based on knowledge-reasoning and
online statistics as much as it can.

Yao Ziyuan

[yaoz...@gmail.com]

An expert-made xword could contain search criteria that effectively
exclude off-topic results, making the end user less disturbed by such
results.

Yao Ziyuan

[yaoz...@gmail.com]

Should an xword be designed in light of experience with searching a
particular document database? If yes, such xwords are called "database
specific", "database oriented" or "database optimized". If no, such
xwords are called "universally applicable".

Both types can be developed for different purposes.

Yao Ziyuan