WO2012069087A1

WO2012069087A1 - Automatic text generation

Info

Publication number: WO2012069087A1
Application number: PCT/EP2010/068246
Authority: WO
Inventors: Jürgen WEWER
Original assignee: Mediasuccess Gmbh
Priority date: 2010-11-25
Filing date: 2010-11-25
Publication date: 2012-05-31

Abstract

The invention proposes a Method for generating a plurality of sentences, a sentence comprising a keyword and an associated feature. In a first step a keyword is Received (100). After having received (200) a plurality of basic sentences, each basic sentence comprising a first indication where a Keyword shall be placed, within said basic sentences said first indication is substituted (300) by said received keyword, thereby generating a plurality of amended sentence. The plurality of basic sentences is provisioned in a further step (700).

Description

Automatic Text Generation

In numerous applications texts are generated for different purposes. Namely in the fields of advertising and Public Relations, texts relating to different issues are composed.

In many fields the information contained in these texts is intended for use by numerous recipients.

However, due to the fact that these texts may be again distributed and published in the internet, the number of texts being essentially the same is growing.

To illustrate the problem, an example will be detailed in the following.

A company likes to provide a press release on a new product. In the past a single press release was produced and afterwards distributed to numerous recipients either by Email, FAX, Mail or by publishing on the Internet.

This leads to a situation where a user seeking for information on said particular product will find the exact information wherever the user seeks. Since users are getting annoyed by these repetitions, users tend to attribute a negative marking to the company or their products.

Said problem arises also in other areas where a recipient receives or retrieves identical information by various source paths.

In order to eliminate the problem the invention proposes a new method for generating a plurality of sentences, a sentence comprising a keyword and an associated feature. Within said method keywords are received and a plurality of basic sentences is received. Each basic sentence comprises a first indication where a Keyword shall be placed. In a further step within said basic sentences said first indication is substituted by said received keyword. In doing so a plurality of amended sentence is generated. Said plurality of amended sentences is provided afterwards.

In an embodiment of the invention, a basic sentence further comprises a second indication where an associated feature shall be placed. Within said embodiment a feature associated to the keyword is received and within said basic sentence said second indication is substituted by said received associated feature.

In a further embodiment of the invention, the plurality of amended sentences is forming a text. In yet another embodiment of said further embodiment of the invention, a desired amount of texts is generated by performing the method steps of said further embodiment of the invention a number of times corresponding to said plurality, wherein the basic sentences of said desired amount of basic sentences are differing one another at least in part.

In a still further embodiment of said yet another embodiment of the invention, the basic sentences pertaining to a keyword and/or an associated feature are of differing length.

According to a further embodiment of the invention the basic sentences provide at least in part for a different prominence with respect to different texts, the prominence being indicative for the number of words within a sentence and the position of a particular word.

According to yet another embodiment of the invention the basic sentences pertaining to an associated feature provide at least in part for a different density with respect to different texts, the density being indicative for the presence of a word within a sentence.

In yet a further embodiment of the invention, the method further comprises a step of substituting at least a word of said basic sentence, whereby said word is not a keyword or an associated feature being subject to substitution in otherwise. In still a further embodiment of the invention, the step of substituting comprises a step of checking whether the step is performed the first time on not, and if the step is performed the first time a word within said basic sentence is substituted.

According to yet another embodiment of the invention, the method comprises also the step of provisioning (1000) analytic results based on the provided amended sentences

Furthermore, the invention is embodied in an apparatus adapted to perform a method according to the invention.

Additionally, the invention is embodied in a Computer-Program-Product allowing for enabling a computer to perform a method according to the invention.

The invention will now be described in further detail with reference to the figures whereby

Figure 1 shows a schematic overall flowchart according to several embodiments of the invention,

Figure 2 shows a schematic detail flowchart according to several embodiments of the invention, and

Figure 3 an exemplary apparatus adopted for performing a method according to the invention. Underlying the invention is the finding that a text of n sentences may be assembled in number of ways. The number of different texts which may be assembled is n! which can be easily shown:

Assume: 10 Sentences S {81 ,82,83,84,85,86,87,88,89,810} being arranged in different texts T

In a first round the location of sentence S1 is altered within the series, see indication in bold face.

T1 = (S1 ,S2,S3,S4,S5,S6,S7,S8,S9,S10)

T2 = (S2,S1 ,S3,S4,S5,S6,S7,S8,S9,S10)

T3 = (S3,S2,S1 ,S4,S5,S6,S7,S8,S9,S10)

T4 = (S4,S2,S3,S1 ,S5,S6,S7,S8,S9,S10)

T5 = (S5,S2,S3,S4,S1 ,S6,S7,S8,S9,S10)

T6 = (S6,S2,S3,S4,S5,S1 ,S7,S8,S9,S10)

T7 = (S7,S2,S3,S4,S5,S6,S1 ,S8,S9,S10)

T8 = (S8,S2,S3,S4,S5,S6,S7,S1 ,S9,S10)

T9 = (S9,S2,S3,S4,S5,S6,S7,S8,S1 ,S10)

T10 = (S10,S2,S3,S4,S5,S6,S7,S8,S9,S1 )

In doing so 10 texts of differing order of the sentences are generated.

In a next round the text starts with Sentence S2, while the position of S1 is altered again:

T1 1 = (S2,S1 ,S3,S4,S5,S6,S7,S8,S9,S10)

T12 = (S2,S3,S1 ,S4,S5,S6,S7,S8,S9,S10)

T13 = (S2,S4,S3,S1 ,S5,S6,S7,S8,S9,S10)

T14 = (S2,S5,S3,S4,S1 ,S6,S7,S8,S9,S10)

T15 = (S2,S6,S3,S4,S5,S1 ,S7,S8,S9,S10)

T16 = (S2,S7,S3,S4,S5,S6,S1 ,S8,S9,S10)

T17 = (S2,S8,S3,S4,S5,S6,S7,S1 ,S9,S10)

T18 = (S2,S9,S3,S4,S5,S6,S7,S8,S1 ,S10)

T19 = (S2,S10,S3,S4,S5,S6,S7,S8,S9,S1 )

In doing so 9 further texts of differing order of the sentences are generated. This process can be continued in the same manner, i.e. the next series starts with sentence S3 while the position of S1 is altered. In doing so 8 further texts of differing order of the sentences are generated

To present this finding on a short example, assume a text composed of n=3 sentences S1 , S2, S3. Than the number of different texts is n!=6=1 ^*2^*3.

T1 = (S1.S2.S3)

T2 = (S2.S1.S3)

T3 = (S3.S2.S1 )

T4 = (S1 .S3.S2)

T5 = (S2.S3.S1 )

T6 = (S3.S1 .S2)

However, these sentences are different with respect to their prominence. The prominence is indicative for the number of words within a sentence and the position of a particular word. However, these texts produced do not provide for a different density. The density is indicative for the presence of a word within a sentence.

This may lead to the problem that although such a toggled text may be arranged in different order the sentences as such do not change which may still be negatively experienced a recipient.

To overcome this drawback each of said n sentences may have n different variants.

Assume the following: A sentence Si may be available in the variants Si1 ,...Sin. Now if sentence Si is used on a position k, where 1 <=k<=n, then variant Sik is taken.

Example:

T1 = (S11 ,S22,S33,S44,S55,S66,S77,S88,S99,S1010)

T2 = (S21 ,S12,S33,S44,S55,S66,S77,S88,S99,S1010)

T3 = (S31 ,S22,S13,S44,S55,S66,S77,S88,S99,S1010)

T4 = (S41 ,S22,S33,S14,S55,S66,S77,S88,S99,S1010)

T5 = (S51 ,S22,S33,S44,S15,S66,S77,S88,S99,S1010)

T6 = (S61 ,S22,S33,S44,S55,S16,S77,S88,S99,S1010)

T7 = (S71 ,S22,S33,S44,S55,S66,S17,S88,S99,S1010)

T8 = (S81 ,S22,S33,S44,S55,S66,S77,S18,S99,S1010) T9 = (891 ,822,833,844,855,866,877,888,519,81010)

T10 = (S101 ,S22,S33,S44,S55,S66,S77,S88,S99,S110)

In doing so, it is enabled that n^*n Sentences allow for n! different texts which are differing both in density and prominence.

Taking this into account when looking at optimizing search engine results, per keyword a typical number of 20 WebPages per Website are needed which shall contain the keyword in a density of about 4 % and a prominence of a certain order, e.g. x %.

If one is using n=8 sentences, each having 8 variants, than a total of 40320 different texts may be generated which may allow for optimizing 40320/20=2016 Websites. In other words, a keyword may be sold 2016 times and each client still has 20 individual and optimized WebPages for the keyword.

In the following examples are described for generating 8^*8 sentences for a particular keyword.

Each Sentence is composed of approximately 25 words. Each sentence shall comprise the keyword. Hence, a density of the keyword of about 4 % is achieved.

In the following it is assumed that the keyword pertains to either

• Behavior (properties/actions), or

• Structure (Product)

Depending to the area of interest the text generation may be performed by different techniques:

I. Behavior (properties/actions)

For example, if the keyword is describing an action, than action descriptions matching the keyword are generated. Standard content is prescribed so that the 64 sentences offer different content each. This is achieved in two dimensions.

In a first dimension matching descriptions of circumstances are selected. Assume a keyword such as "demolition". Than a set of matching descriptions of circumstances may be composed of:

1 . performing 2. planning

3. checking

4. execution

5. organization

6. processing

7. implementing

8. realizing

Thereby 8 basic sentences are resulted, comprising the following words:

51 = (Demolition, Performing)

52 = (Demolition, Planning)

53 = (Demolition, checking)

54 = (Demolition, Execution)

55 = (Demolition, Organization)

56 = (Demolition, Processing)

57 = (Demolition, Implementing)

58 = (Demolition, Realizing)

A second dimension is realized by the combination with sub-issues (before start, on start, during, frequent, important, in parallel, at the end, after ending) which may further define place and or time or the like.

To visualize this issue, a matrix is set-up, however showing only some entries for ease of understanding:

n

frequent safegua inquire

rding

importa

nt

In

parallel

At the disposal offer

end

After order debriefi ending ng

Hence, a set of 64 sentences having different thematic content may be construed:

51 1 = (Demolition, Performing, before start, Site inspection)

512 = (Demolition, Performing, on start, barriers)

513 = (Demolition, Performing, during, dust formation)

521 = (Demolition, Planning, before start, contact person)

522 = (Demolition, Planning, on start, site conference)

S88 = (Demolition, Realizing, after ending, Debriefing)

It is rather easy to compose sentences thereof.

Example: Keyword:„Search Engine Optimization "

Here the matrix may be as follows:

Before d n of chains Tag structur n of sub- g start of Analysi targets e content domains Websit s manage e

ment

system

b. On Key wo r Region Implemen Descrip IP- TYP03 Content Home start of d al ting of tion- Address Planning page allocati concep multilingu Tag es

on t alism

c. Checki Extern Semantic Headin Own Domain Avoiding Duplica

During ng of al sentence g-Tag server age JavaScri ted

Positio linkage s pt content n

d. Content Interna Consiste Alt-Tag Linux Alexa Title tag Web

Freque concept 1 ncy of Server length index ntly linkage chains

e. Measur Page Pictures Link- Window dmoz Descripti Linkag import ing analysi Tag s Server on tag e ant density s length originat ing from the home page f. in Measur Compe Ajax Bold- News- Page Visitor Conver Paralle ing titor Tag concept Impressi sion 1 promin analysi ons rate ence s

g. At Google Offer Frameles URL Position Log files Keyword Backup the Analytic s reportin position end of s g System h. Visitor Order Number Speaki Ranking Admin Editorial Google

After increas of ng URL rights office -Maps ending e outbound

of links

Table 1

Hence, again a set of sentences having different thematic content may be construed:

1 .a: Before start of Performing Search Engine Optimization it is necessary to generate a Keyword Analysis. There, the most important keywords having the highest search volume are identified.

1 .b: On start of Performing Search Engine Optimization a Keyword allocation onto possible sub-pages of the Website is necessary. It is determined on which sub-pages which keywords are to be represented.

1 .c: A Checking of Position is an important task during Performing Search Engine Optimization. It is performed with respect to the most important keywords.

1 .d: Frequently Performing Search Engine Optimization is accompanied by preparation of a content concept. I.e. it is determined which content the keywords are used.

1 .e: Measuring density is important during Performing Search Engine Optimization . The density of a word is calculated as ratio of the presence (i.e. the number of occurrence) of a word versus the total number of words.

1 .f: Measuring of Prominence is performed in parallel to Performing Search Engine Optimization. It takes into account the actual position of a word in relation to its density.

1 .g: At the end of Performing Search Engine Optimization Google Analytics shall be installed on the website of the client in order to measure the number of Visitors.

1 . h: After ending of Performing Search Engine Optimization a strong Visitor increase may be anticipated.

2. a: Before start of Planning Search Engine Optimization the position of targets shall be defined. For each Keyword a certain page area shall be defined.

2.b: On start of Planning Search Engine Optimization a Regional concept is developed. I.e. it is determined in which city or region a client shall be found.

2.c: During Planning Search Engine Optimization a concept for External linkage shall be developed. I.e. it is determined towards which websites having similar content links may be positioned. 2.d: Frequently on Planning Search Engine Optimization also the Internal linkage is developed. I.e. it is determined which menu structure is mapped onto the different tiers and where exist deeper links.

2.e: An individual Page analysis is important while Planning Search Engine Optimization as well as a cumulative analysis of all pages of a Website.

2.f: Competitor analysis is performed in parallel to Planning Search Engine Optimization. I.e. it is determined how many indexed pages are originating from competitors and which is their page rank.

2.g: At the end of Planning Search Engine Optimization a concrete offer shall be present. I.e. it shall be clear which service may be offered at a certain pricing.

2. h: After ending of Planning Search Engine Optimization, if everything went well, an order is to be expected.

3. a: Before start of checking Search Engine Optimization it shall be placed emphasis on the development of a good strategy for Link chains. The Link-Text, the Link itself and the page which is linked shall comprise the same keywords.

3.b: On start Checking Search Engine Optimization it is necessary to pay attention to Implementing multilingualism.

3.c: During Checking Search Engine Optimization a constant control is necessary that alls Tags are filled by semantic sentences being meaningful. A mere series of words is of no particular relevance any more.

3.d: Frequently Checking Search Engine Optimization neglects consistency of Link chains. It is of particular interest to keep consistency.

3.e: During Checking Search Engine Optimization it is important to implement pictures on a Webpage having implemented respectively filed Alt-Tags. Preferably these Alt-Tags shall be filed by complete sentences.

3.f: In parallel to Checking Search Engine Optimization it is necessary to take into account that using AJAX leads to a situation where particular content is loaded on first hit of a Webpage.

3.g: At the end of Checking Search Engine Optimization it shall be verified that the Website is also implemented frameless. The content of frames is not attributed to the original domain. 3. h: After ending of Checking Search Engine Optimization the number of outbound links shall be checked. It is a Tendency that that this leads to a decrease of relevance. Relevance is attributed in a linear manner to the outbound links.

4. a: On Execution of Search Engine Optimization the title-tag has an important function. It is typically displayed within the first line of a search hit and shall display a meaningful sentence of about 80 chars.

4.b: On start of Execution Search Engine Optimization the Description- Tag is in focus. Content is displayed in general below the title within the search hits and shall comprise about 180 chars and a sentence having a subject, verb and object. The most important keyword shall be placed at the beginning of the sentence,

4.c: During Execution of Search Engine Optimization Heading-Tags shall be used. Content within headings have an outstanding weight for the complete text.

4.d: Frequently the Alt-Tag is neglected during Execution of Search Engine Optimization. Nevertheless, Pictures shall not be missed and the Alt-Tags shall comprise a description of the picture content or the impression of the picture. These Alt-Tags shall contain the important keywords as well.

4.e: Link-Tags are Important for Execution of Search Engine Optimization. Links constitute the network of the internet. Without them - practically speaking - there would be no Internet. All Links shall have a Link-Text, which is consistent to the link itself.

4.f: Execution of Search Engine Optimization shall in parallel take the Bold-Tag into account. Issues indicated in bold face or otherwise being highlighted - which is a particular aim of a Markup Language - offer an outstanding weight for the complete text.

4.g: At the end of Execution Search Engine Optimization URLs shall be checked once again. It is important to understand that the URL is displayed already as third element of a search hit.

4. h: After ending Execution of Search Engine Optimization Speaking URL are to be implemented. In particular when using Content Management Systems special extensions are to be configured.

5. a: Before start of Search Engine Optimization Organization it is necessary to ensure a suitable Hosting structure. Many Domains having a common IP-Address may allow for the conclusion that the Website is missing wholeheartedness 5.b: On start of Search Engine Optimization Organization a Hosting package shall be selected that does not share the IP-Address or at least minimizes the number of sharing partners.

5.c: During Search Engine Optimization Organization for larger projects an own server shall be envisaged. Such own servers may be ordered at relative moderate prices and offers an own IP-Address.

5.d: Search Engine Optimization Organization frequently makes use of a Linux-Server. They are known to be stable and are widely used and they offer by far the most applications.

5.e: During Search Engine Optimization Organization it is important that if one is opting for a Windows-Server to ensure having a well skilled software engineer at hand which is familiar with a windows-Server.

5.f:ln Parallel to Search Engine Optimization Organization a News-concept shall be developed which contributes actual content for the respective issue on a constant basis.

5.g: At the end of Search Engine Optimization Organization a Positions Reporting shall be implemented, which on a daily basis shows the own position of the WebPages for the important keywords.

5. h: After ending of Search Engine Optimization Organization an improved ranking shall be achieved which shall increase traffic towards the Website significantly.

6. a: Before start of Processing Search Engine Optimization a suitable Content Management Systems shall be selected.

6.b:On start of Processing Search Engine Optimization the selection of Content Management System often leads to TYP03. This system is wide spread, stable, freely available and offers by far the most applications.

6.c: During Processing Search Engine Optimization the Domain age may be of concern. Old Domains typically are held to be more reliable by search engines.

6.d: While Processing Search Engine Optimization frequently Alexa is taken into account.. This system offers a certain reference for the traffic of a domain and the degree of external linkage.

6.e: While Processing Search Engine Optimization it is important to accomplish an entry into the most important Web index, i.e. DMOZ. In doing so, the reliability of the Website as attributed by the Search engines increases since they offer an expertise by an editorial office. 6.f: In Parallel to Processing Search Engine Optimization the development of the Page Impressions shall be in focus. It shows how often the Website is displayed by visitors.

6.g: At the end of Processing Search Engine Optimization the Log files shall be of concern. What shall be logged and analyzed how and when. These parameters are to be taken into account when programming the server.

6. h: After ending of Processing Search Engine Optimization the respective Admin Rights are to be attributed both on the Server and the Content Management System.

7. a: Before start of Implementing Search Engine Optimization typically the structure of the presence shall be decided, i.e. whether one is using sub-domains or subfolders. It is to be understood that sub-domains are treated by the search engines as if they would be single domains.

7.b: On start of Implementing Search Engine Optimization a Planning of content is necessary. I.e.. it needs to be determined which content shall be published on a particular Webpage.

7.c: During Implementing Search Engine Optimization it is important Avoiding JavaScript. JavaScript and the results provided by JavaScript-requests are not visible for search engines.

7.d: Frequently it happens that while Implementing Search Engine Optimization title Tag length is not considered. A Title Tag shall not exceed 80 chars.

7.e: Selecting a sufficient Description tag length is Important while Implementing Search Engine Optimization. The Description tag length shall be around 160 chars.

7.f: In Parallel to Implementing Search Engine Optimization it is important to track changes of the Visitors. They shall increase while implementing the measures.

7.g: At the end of Implementing Search Engine Optimization it is important to select a good Keyword position within the texts. Keywords shall be placed preferably at the beginning of a text and there shall be no repetitions at the end which may ruin the presence of a keyword.

7. h: After ending of Implementing Search Engine Optimization an Editorial office shall be setup which cares of the Website in a continuous manner and which supplies actual content.

8. a: Before start of Realizing Search Engine Optimization a thorough analysis of the existing Website is performed. Where the important keywords are and which density and prominence do they offer. 8.b: On start of Realizing Search Engine Optimization the Home page shall be of particular concern since the Homepage has by far the most relevance.

8.c: During Realizing Search Engine Optimization Duplicated Content shall be avoided, since copied content is penalized.

8.d: While Realizing Search Engine Optimization it is frequently neglected to register at major web indexes. By registering with DMOZ a lot may be achieved by spending little efforts.

8.e: By providing a number of outbound links of important keywords while Realizing Search Engine Optimization the relevance may be increased.

8.f: In Parallel to Realizing Search Engine Optimization Conversion Rate shall be taken into account since it is the goal not only to generate traffic but traffic which turns into clients.

8.g: At the of Realizing Search Engine Optimization a suitable Backup-System shall be selected, in order not to be dependent on Hoster's Backup in case of Server Failure.

8.h: After ending of Realizing Search Engine Optimization it is a necessity to create an entry for Google-Maps. Hereby it is achieved that the client is already present in the local search hits.

II. Structure (Product):

The keyword pertains to Structure, i.e. it describes a product:

In this case the phrase texts may be even parameterized. Thereby a number of sets (classes) of keywords may be generated. For those keywords sentences may be generated comprising different features while using like basic sentences thereby providing different texts.

Example: Keyword =„VW Golf"

The Keyword pertains to the category car. For each car a number of different features may be named, such as HP, top speed, Acceleration, Weight, Consumption, Trunk size, Payload, Trailer load.

For each feature a number of different basic sentences are determined having the feature as parameter. I.e. in the following it is assumed that for each of the above mentioned 8 features 8 basic sentences are determined.

Example: Keyword =„VM Golf", feature =„HP" 1 . HP 2. Top 3. 4. 5. 6. 7. 8.

speed accele Weig Consu Trunk Paylo Traile ration ht mption size ad r load a The <keyword>

Specific posses an

ation excellent engine

having <HP> HP.

b. Having <HP> HP

Power the <keyword>

may easily be

classified as a

medium-sized

vehicle.

c. Fun of driving may

Dimens be expected when

ions driving the m

<keyword> having

an engine

performance of

<HP> HP.

d. The <keyword>

Weight offers <HP> HP,

which provides a

sufficient power

within its class.

e. The <keyword> is

Equipm powered by <HP>

ent HP allowing for a

speedy travelling .

f. In today's traffic

Format the <keyword>

may easily flow

having <HP> HP. g. Data The Weight of the

<keyword> may

easily be driven

taking into account

<HP> HP, allowing

for sufficient

reso u rces eve n

when being driven

in alpine areas.

h. Having <HP>HP

Volume allows the

<keyword> to be

driven in an

enjoyable manner

while not showing

any deficit.

Table 2

If this matrix is filled, 64 sentences for a keyword having each a parameter are provided.

In more general, for any product, which may not be described in features as above, it still is possible to describe the product by standardized properties or actions.

Example: Keyword„Website Tool":

Table 3 This approach is suitable for products/objects/persons which are unique or where it can be assumed that they are rare.

1 .a The Website Tool promises on start a thorough Analysis.

1 .b Using a Website Tool promises frequently a mathematical approach for optimization.

1 .c During Operation using a Website Tool promises a thorough Source code Analysis.

1 .d At the end a Website Tool promises a Ranking Prognosis.

2. a On start The Website Tool operates on basis of URL-Addresses.

2.b Frequently the Website Tool operates on a dedicated Web Server.

2.c During use the Website Tool operates on all Subpages.

2. d At the end a Website Tool operates to achieve an objective result.

3. a A good Website Tool analyses On start the exact amount of words.

3.b Frequently a Website Tool analyzes the Title-Tag of the HTML-header.

3.c During operation a Website Tool analyzes the Description Tag.

3. d At the end a Website Tool analyzes the Link structure

4. a On start a Website Tool allows for Word counting.

4.b Frequently a Website Tool allows for a Calculation of density.

4.c During operation a Website Tool allows for Calculation of prominence.

4.d At the end a Website Tool allows for a Cloud of words, allowing for visualizing the ranking of words.

Alternatively also features for the product„Website Tool" <WS> may be determined:

extraordinary frequently a importance for <EG>.

displaying

search results.

Subject Taking into For a thorough Frequently one If At the end of account the Analysis the needs to wait the analysis by particular <WS> counts only fractions of the <WS> a importance of <ZE>. seconds for the <EG> is <Tag> the calculation of achieved one <WS> analyzes <BE> when may be said issue in a using a <WS>. positively special manner. surprised.

Object The <WS> To achieve an A reliable Result A well working analyzes On objective result is attained by a <WS> ends start the <Tag> the <WS> <WS> by operation on first hand. counts <ZE> calculating yielding a <EG>.

<BE>.

At the end At the end, any The <WS> A <WS> may be A sophisticated result achieved measures for an graded well <WS> yields a by a <WS> is objective when <BE> is <EG> At the depending on comparison calculated in an end

Analysis of <ZE> appropriate

<Tag> manner..

Table 4

The Product and its feature:

1 . Website Tool WOM-Title {(Tag, Title-Tag), (ZE, words),(BE, Density), (EG, Cloud of words)}

2. Website Tool WOM-Description {(Tag, Description-Tag),(ZE, words),(BE, Prominence), (EG, Ranking Prognosis)}

3. Website Tool WOM-Body {(Tag, Body-Tag),(ZE, words),(BE, Density),(EG, Cloud of words)} 4. Website Tool WOM-Link {(Tag, Link-Tag),(ZE, words),(BE, Prominence),(EG, Ranking Prognosis)}

5. Website Tool WOM-Bold {(Tag, Bold-Tag),(ZE, words),(BE, Density),(EG, Cloud of words)}

Per product by the above method it is possible to generate 24 different Texts having each 4 sentences may be generated. Taking into account the above mentioned 5 products, i.e. Website Tool WOM-Title, Website Tool WOM-Description, Website Tool WOM-Body, Website Tool WOM-Link and Website Tool WOM-Bold, a total of 120 different Websites may be generated.

Category, Parameter:

Both algorithms relating to keywords pertaining to the above mentioned areas may be generalized by adding category and parameter. Each keyword is attributed to a category and each category is attributed to a 2 dimensional list or matrix of basic sentences. Each basic sentence may comprise parameter having each a unique ID. In a similar manner each keyword may have one or more attributed Parameter-IDs having corresponding values.

Keyword: -> Category

Category: -> {Basic sentences, i,j}

Basic sentence(ij) comprises {Parameter n, Parameter m}

Keyword: -> {(Parameter n, Value Parameter n), (Parameter m, Value Parameter m)}

Example:

1 . Product: Keyword„VW Golf"

Category: PKW

Category„PKW" ->„Basic sentences see Table 2, VW Golf

Basic sentence (1 ,2) comprises {Parameter 1 1 , Parameter 12}

Parameter 1 1 : keyword; Parameter 12: HP

„VW Golf -> {(keyword,„VW Golf), (HP, 54)} 2. Action: Keyword„Search Engine Optimization "

Category: Search Engine Optimization

Category„Search Engine Optimization " ->„Basic sentences see Table 1 , Search Engine Optimization "

Basic sentence (i,j) comprises {} - empty set - no parameter

Keyword„Search Engine Optimization " -> {} - empty set - no Parameter

Unique Attribution of pages to the clients:

A first client will receive e.g. the first 20 pages, the first page comprises the sentences S1 1 ,S22,S33, ....,S88. The second page comprises the sentences S21.S12.S33, ...,S88

A second client will receive another 20 pages, ... . The exact attribution is of no further concern and may be accomplished by existing algorithms.

Finally, the newly generated and optimized WebPages shall be integrated into existing Web Site-Structure.

The WebPages optimized with respect to the keywords may be integrated into clients Homepage by a so-called glossary. Such a glossary may be of dynamic nature and typically is of a database type and following links will be generated:

For each category the names of the dimensions of the basic sentences are stored along.

Example 1 : Category: Search Engine Optimization, Keyword: Search Engine Optimization

Dimension 1 : {Performing, Planning, Surveillance, Execution, Organization, Processing, Implementing, Realizing}

Dimension 2: {Before start, on start, During, Frequently, Important, In Parallel to, at the end, After ending}

Example 2: Category: PKW, Keyword: VW Golf

Dimension 1 : {HP, top speed, Acceleration, Weight, Consumption, Trunk size, Payload, Trailer load } Dimension 2: {Specification, Power, Dimensions, Weight, Equipment, Format, Data, Volume}

For each Webpage the Link is generated using the Keyword and both names of the dimensions of the associated category of two consecutive sentences.

Thereby 8 x 56 = 448 different Links may be generated. Hence, per client 448 different links may be generated for a single keyword.

The 8 x 56 Links are generated as follows.

The first 56 links:

The first 8 Links are generated using Names of dimension of the 1^st sentence and the Names of dimension of the 2^nd sentence.

The next 8 links are generated using Names of dimension of the 1^st sentence+ Names of dimension of the 3^rd sentence. This process is continued until the combination of the Names of dimension of the 1 ^st sentence and the Names of dimension of the 8^th sentence is reached.

The next 56 links:

The first 8 links are generated using Names of dimension of the 2^nd sentence and the Names of dimension of the 3^rd sentence.

The next 8 links are generated using Names of dimension of the 2^nd and the Names of dimension of the 4^th sentence. This process is continued until the combination of the Names of dimension of the 2^nd sentence and the Names of dimension of the 1^st sentence is reached. Here, it is to be recognized that after the combination having the 8^th sentence, the process continues at the 1^st sentence.

The process is repeated for the next 56 links until the 8^th set of 56 links is generated as follows:

The first 8 Links are generated using Names of dimension of the 8^th sentence and the Names of dimension of the 1 ^st sentence. This process is continued until the combination of the Names of dimension of the 8^th sentence and the Names of dimension of the 7^th sentence is reached. The respective keyword is added to the link. Preferably the keyword is put at the beginning of the link.

Example:

1 . Search Engine Optimization (Performing, before start) (Planning, On start)

2. Search Engine Optimization (Performing, On start) (Planning, On start)

3. Search Engine Optimization (Performing, during) (Planning, On start)

4. Search Engine Optimization (Performing, frequently) (Planning, On start)

5. Search Engine Optimization (Performing, important) (Planning, On start)

6. Search Engine Optimization (Performing, in parallel to) (Planning, On start)

7. Search Engine Optimization (Performing, At the end ) (Planning, On start)

8. Search Engine Optimization (Performing, after ending) (Planning, On start)

9. Search Engine Optimization (Performing, before start) (Surveillance, On start)

10. Search Engine Optimization (Performing, On start) (Surveillance, On start)

1 1 . Search Engine Optimization (Performing, during) (Surveillance, On start)

12. Search Engine Optimization (Performing, frequently) (Surveillance, On start)

13. Search Engine Optimization (Performing, important) (Surveillance, On start)

14. Search Engine Optimization (Performing, in parallel to ) (Surveillance, On start)

15. Search Engine Optimization (Performing, At the end ) (Surveillance, On start)

16. Search Engine Optimization (Performing, after ending) (Surveillance, On start)

Realizing:

It is beneficial to store the pages per keyword for each client individually. Furthermore, it is preferred that the integration into a glossary shall be effected in a controlled manner, e.g. only a certain number of optimized WebPages shall be made available at one time. In doing so, the effect may of the optimization may be tracked more easily since it is stretched over time and the optimization may take new findings immediately into account. Also visitors experience may be negatively affected if a Website is migrated in an abrupt manner. Having now described the background and given an glimpse on the operational aspects and possible fields of usage embodiments of the invention will be further detailed in the following.

A method as shown in exemplary flowchart of Figure 1 which may be performed on a computer may be divided into several steps. It is to be understood that unless a particular sequence is necessitated in that particular information is gathered before being used the steps itself may be arranged in any appropriate manner. It may even be possible to perform method steps in parallel.

Once said method is started in a first step 100 a keyword is received. In another step 200 a plurality of basic sentences is received. Each basic sentence comprises a first indication where a Keyword shall be placed. Obviously said basic sentences may also comprise further indications for other keywords.

It is also to be understood, that receiving may encompass a receiving as an input via a keyboard, a process of loading keywords from a data carrier or receiving from some network site. Receiving information may also encompass a process of actively retrieving information and should therefore be understood in the broadest manner.

Once basic sentences and keywords are received, within said basic sentences said first indication is substituted in a step 300 by said received keyword <keyword>, thereby generating a plurality of amended sentence.

In a most basic embodiment said generated amended sentences are provisioned in a step 700.

It is to be understood that provisioning may encompass a storing on a data carrier such as a hard disk (HDD), in memory (RAM), on another network site or by displaying on a computer or providing a print out. Provisioning of information may also encompass a process of passively making available for retrieval and should therefore be understood in the broadest manner.

Once the provisioning is accomplished, the basic method ends.

In an enhanced embodiment, the method allows for basic sentences further comprising a second indication where an associated feature shall be placed.

The method now also comprises a step of receiving 400 a feature associated to the keyword.

Consequently the method also comprises a step of substituting 500 within said basic sentence said second indication by said received associated feature. Obviously, the step of receiving 400 a feature associated to the keyword and / or the step of receiving 100 keywords and / or the step of receiving 200 basic sentences may be performed sequentially or in parallel.

Also the steps of substituting 300 within said basic sentences said first indication and the step of substituting 500 within said basic sentence said second indication by said received associated feature may be performed sequentially or in parallel. The only dependence they have is that the appropriate information needs to be available.

Once the sentences are amended they may be provisioned in a step 800 as described before.

In a still further embodiment 8 a step of substituting 700 at least a word of said basic sentence is provided, whereby said word is not a keyword or an associated feature being subject to substitution in otherwise.

In doing so, a still further degree of uniqueness of individual sentences generated by the automatic text generation may be achieved.

Suppose the following:

A text may be composed of several basic sentences. In the following we will turn to an example pertaining to Mobile Phones which may have been generated by an automatic text generation as described before:

„Having a Nokia 5230 Navi enables Navigation services free of charge and virtually any time by its included GPS receiver. For those taking pictures spontaneously the Nokia 5230 Navi provides a 2,0 Megapixel-camera. In the black edition the Nokia 5230 Navi represents an unrivaled synthesis of design and function. By providing a Bluetooth interface a fast data transfer is enabled. Enjoy Music downloaded from the Nokia Music Store by listening with the Stereo Headset of the Nokia 5230 Navi. For ease of use the Nokia 5230 Navi is packaged with a car-mount. Having a Touch- Screen the Nokia 5230 Navi allows its user to experience a modern user interface. By having a memory card of 2GB included, the Nokia 5230 Navi may even store huge data amounts reliably."

This text may have been generated as before by use of an 8^*8 matrix which may lead to 40320 texts which may have different density and prominence.

Each one of these sentences may be individualized by using synonyms. In doing so, one may arrive at a plurality of text where each and every sentence is present only once.

Let us take the first sentence of the above reproduced text as an example:

"Having a Nokia 5230 Navi enables Navigation services free of charge and virtually any time by its included GPS receiver."

In the methods as described above this sentence would occur at different positions within each 8th text, i.e. assuming all text are generated (40320) than the sentence would occur in total 5040 times. Introducing synonyms a plurality of different sentences may be generated, whereby the plurality is defined by the number of synonyms available for all words other than those words which are to be substituted anyhow.

Synonyms may be derived from databases providing such dictionaries or may even be derived from on-line services. Examples for such online services are the German Website of the University of Leipzig (http://wortschatz.uni-leipzig.de/) providing German synonyms or Merriam-Webster-Online services (http://www.merriam-webster.com/) providing for English synonyms or for different languages LEO service (http://dict.leo.org/).

The basic sentence underlying the above sentence may look as follows:

"Having a <keyword> enables Navigation services free of charge and virtually any time by its included <feature>."

Now the Synonyms are determined:

"Having":

Holding, maintaining, including -> 3 Synonyms

"enables"

Provides, makes possible, allows for -> 3 Synonyms

"Navigation Service"

Determination of course, Routing -> 2 Synonyms

"Service"

Supply, Aid -> 2 Synonyms

"free of charge"

at no charge, gratis, gratuitous, toll-free, nonpaid, for free, freebie- > 7 Synonyms "virtually"

Almost, nearly, close-to, well-nigh, practically > 5 Synonyms

"any time"

At any time, at all times, all-times -> 2 Synonyms

"included"

Contained, comprised, inserted, embedded -> 4 Synonyms

Hence having these synonyms at hand, one may easily generate 3^*3^*2^*2^*7^*5^*2^*4= 10080 amended basic sentences. Hence, in total 10081 different sentences are yielded.

Now having described this arrangement, it is clear that the above captioned method may be used in addition or as an alternative to the methods described previously. I.e. one may even use the above mentioned method of substituting synonyms and operate it on each sentence of a matrix.

In doing so, one is able to generate a high number of different texts.

Taking a n^*n matrix symbolized as t(i,j) where i,j are elements of the series 1 ,2,...n and v(i,j) representing the number of variants of a sentence after having been subject to the process of substituting synonyms, than the maximum amount of different Texts M may be derived as follows:

M = min {s(1 ), s(n)}, whereby s(i) = v(i1 ) + ... + v(in).

As stated the above method of substituting synonyms may be effected alone or in combination with other described methods. Hence, the exact location within the process may be chosen appropriately. The method step 700 may e.g. performed immediately after having received a basic sentence or may even performed after having substituted keywords and or associated features.

In more detail, the process of substituting 700 may comprises a step of checking 710 whether the step is performed the first time on not, and if the step 700 is performed the first time a word within said basic sentence is substituted 720. For this purpose one may check a counter N which counts the number of cycles already performed by the method. It is to be understood that the exact implementation may also differ in that it is checked beforehand whether the condition is fulfilled or not. As such the exact implementation is left open as it is well known in the art how to design appropriate conditioned loops.

In doing so, the basic sentence will be present at least within the first text.

It is particular advantageous if the plurality of amended sentences provided by either method is forming a text. Text is to be understood in a very general manner, i.e. it encompasses any kind of text such as WebPages, Email, Printouts and or audio presentations of text-to- Speech-Converters.

When texts for a plurality of keywords are to be generated the method steps as described before may be performed a number of times with respect to particular keywords. For this reason the flowchart displays a decision step 900 in which it is determined whether the necessary amount of amended sentences is achieved or not. It is to be understood that the exact implementation may also differ in that it is checked beforehand whether the condition is fulfilled or not. As such the exact implementation is left open as it is well known in the art how to design appropriate conditioned loops. For this purpose, in figure 1 it is indicated a condition with respect to a counter N which counts the number of cycles already performed by the method. If the counter reaches a certain threshold M, which may be the maximum number of different sentences, than the check leads to a proceeding to step 1000. Advantageously the basic sentences of said desired amount of basic sentences are differing one another at least in part. However, this is not a necessity.

It is further advantageous if the basic sentences pertaining to a keyword and/or an associated feature are of differing length.

By having basic sentences which have differing length, it is achieved that the generated texts provide for different prominence and / or density with respect to any word, in particular a keyword or an associated feature.

Obviously, as described above in the general section, a further step 1000 may be provisioned which provides analytic results based on the provided amended sentences. These results may pertain to the density and / or prominence of one or more keyword and / or one or more associated feature to allow for controlling the optimization result.

Having detailed the operation, it is readily understandable that the methods as outlined before may be embodied in apparatus such as a computer.

An exemplary apparatus is shown in figure 3. Said apparatus may be equipped with a CPU for operating substituting steps. Furthermore, the apparatus may comprise an I/O unit which allows for interaction by keyboard, by receiving data from external sources such as another internet site and / or for provisioning amended sentences.

Furthermore, the apparatus may be equipped with a memory RAM for storing a program allowing the computer to perform the method steps. Also the memory may allow storing an intermediate result.

Obviously, an apparatus may also be equipped with a data carrier such as a hard drive HDD for storing results and intermediates of the optimization.

Having stated this it is also clear, that the method may be stored on a Computer-Program- Product such as a CD, DVD, a Network Attached Storage, A Download Server, allowing for enabling a computer to perform a method according to the invention.

Without being limited to Webpage Generation, the teaching of the invention may be used for several purposes and is not limited to online publications.

It is noted that in particular in the fields of advertising and Public Relations, texts relating to different issues may be generated in an automated manner thereby allowing for enhanced search results. Furthermore, since the texts are of different style due to the automatic composition the readers experience is positively influenced.

To show the benefits again, we return to the example discussed in the background section and apply to the newly invented automatic text generation. A company likes to provide a press release on a new product. In the past a single press release was produced and afterwards distributed to numerous recipients either by Email, FAX, Mail or by publishing on the Internet.

By using the automatic text generation, a plurality of different texts may be generated. Each text may contain the same information but it is transported in different flavors due to the different amended sentences making up the texts.

Hence, even if the user seeks for information on a particular product and may even find the same underlying information, the experience is different than before, because now the texts are not exact copies of one another but differing in many aspects. Therefore a user will no longer get annoyed by these repetitions, and thereby users tend to attribute a positive marking to the company or their products.

Although particular embodiments have been described, it is to be understood that the invention is not limited thereto but the embodiments served as examples only.

Claims

1 . Method for generating a plurality of sentences, a sentence comprising a keyword and an associated feature, comprising the steps of:

• Receiving (100) a keyword,

• Receiving (200) a plurality of basic sentences, each basic sentence comprising a first indication where a Keyword shall be placed,

• Substituting (300) within said basic sentences said first indication by said received keyword, thereby generating a plurality of amended sentence,

• Providing (800) the plurality of amended sentences.

2. Method according to claim 1 , wherein a basic sentence further comprises a second indication where an associated feature shall be placed, the method further comprising the steps of:

• Receiving (400) a feature associated to the keyword,

• Substituting (500) within said basic sentence said second indication by said received associated feature.

3. Method according to claim 1 or 2, wherein the plurality of amended sentences is forming a text.

4. Method according to claim 3, wherein a desired amount of texts is generated by performing the method steps according to claim 3 a number of times corresponding to said plurality, wherein the basic sentences of said desired amount of basic sentences are differing one another at least in part.

5. Method according to claim 4, wherein the basic sentences pertaining to a keyword and/or an associated feature are of differing length.

6. Method according to claim 4 or 5, wherein the basic sentences provide at least in part for a different prominence with respect to different texts, the prominence being indicative for the number of words within a sentence and the position of a particular word.

7. Method according to claim 4 or 5 or 6, wherein the basic sentences pertaining to an associated feature provide at least in part for a different density with respect to different texts, the density being indicative for the presence of a word within a sentence.

8. Method according to one of claims 3 to 7 further comprising the step of substituting (700) at least a word of said basic sentence, whereby said word is not a keyword or an associated feature being subject to substitution in otherwise.

9. Method according to claim 8, wherein the step of substituting (700) comprises a step of checking (710) whether the step is performed the first time on not, and if the step of substituting (700) is performed the first time a word within said basic sentence is substituted (720).

10. Method according to any one of the proceeding claims, further comprising the step of provisioning (1000) analytic results based on the provided amended sentences.

1 1 . Apparatus adopted for performing a method according to one of claims 1 to 10.

12. Computer-Program-Product allowing for enabling a computer to perform a method according to one of claims 1 to 10.