A DETECTION METHOD OF STAGNATION SYMPTOMS BY USING

PROJECT PROGRESS MODELS GENERATED FROM PROJECT

REPORTS

Satoshi Tsuji, Yoshitmo Ikkai, Masanori Akiyoshi

Graduate School of Information Science and Technology

Osaka University

Suita, Osaka, Japan

Keywords:

Text mining, stagnation symptom, knowledge management, project management, modeling.

Abstract:

The purpose of this research is to extract “stagnation symptoms” from progress reports related to a research

project. A stagnation symptom is deﬁned in a portion where remarkable stagnation is seen during the progress

of a project. Speciﬁcally, according to project managers, stagnation symptoms can be classiﬁed into the

following three kinds: ﬁrst one is a project bottleneck grasped from one document; the second is clariﬁed by

comparing it with the most recent document; and the third is clariﬁed from changes to a working object in a

series of documents. We propose a method of extracting stagnation symptoms using the structural analysis

of a project’s progress. A progress model that is a structural chart to expressing the progress of a project is

generated from documents with label tags, which indicate prior contexts or attributes. This progress model has

the following features: a multilevel layer model using detailed degrees and situation analysis using color, and

relation analysis of these details and basis using color propagation. Stagnation symptoms are automatically

extracted by applying stagnation symptom extraction rules to the progress model. This proposed method

was been applied to a set of real progress reports. It could extract stagnation symptoms that were extracted

manually.

1 INTRODUCTION

Research projects are planned and enacted at enter-

prises and universities every day. In progress of

a projects, progress management greatly inﬂuences

the success of a project(Takemoto, 2005; Sakthivel

and Kalyanaraman, 1993). At this time, an enor-

mous array of varied documents such as business

daily reports, weekly reports, and progress conﬁrma-

tion minutes meeting are drawn up and circulated. To

manage progress, management staff must examine all

these documents(Paul and Khan, 1999; Kloppenborg

and Petrick, 2004). If a project has some problems,

they need to think about appropriate measures. How-

ever, looking for problem that is not necessarily de-

scribed in an enormous number of documents requires

a large amount of labor. Therefore, there is a strong

need to develop a system that automatically extracts

portions of a problem about the progress of a project

from progress reports.

In existing research, a method that accumulates

data, such as similar failure cases in the past, has been

proposed and used for project management(Tsukuda

and Morita, 2004; Y. Uchida and Tatebe, 2005). How-

ever, there is an assumption that data of similar cases

in the past has been accumulated in this research. In

any case, these methods are not effective if there is no

such data.

In our research, a portion where remarkable stagna-

tion is seen during the progress of a project is deﬁned

as a “stagnation symptom”. Stagnation symptoms are

extracted using the following method. First, label tags

that show attribute information such as ID, work con-

tent, and so on are added to each sentence of an input

documents. Next, a progress model that is a structural

chart that expresses the progress of a project is gener-

ated from documents with label tags. Analyzing the

structure of the generated model and applying stagna-

tion symptom extraction rules to the progress model

then extract stagnation symptoms.

2 STAGNATION SYMPTOMS

EXTRACTION

2.1 Stagnation Symptoms

A portion where remarkable stagnation is seen in a

project progress is deﬁned as a “stagnation symp-

165

Tsuji S., Ikkai Y. and Akiyoshi M. (2006).

A DETECTION METHOD OF STAGNATION SYMPTOMS BY USING PROJECT PROGRESS MODELS GENERATED FROM PROJECT REPORTS.

In Proceedings of the First International Conference on Software and Data Technologies, pages 165-169

DOI: 10.5220/0001315401650169

 SciTePress

tom”. Figure 1 shows a typical example in which

project managers consider stagnation symptoms when

reading documents.

1st report

1.Work report

3.Impression

2.Problem in the future

•I want to investigate the

background firmly, and

to prove it.

•Problem of the research

was set.

•Background investigation

•Design of approach

1.Work report

3.Impression

2.Problem in the future

•It is necessary to still

investigate the problem

setting.

•An approach was set.

•There is still an opaque part

though the background was

investigated.

•Examination of approach

•Background is continuously

investigated.

1.Work report

3.Impression

2.Problem in the future

•An approach is not

effective under a present

problem setting.

•Review of problem setting

•Background investigation

•I want to find the condition

that the approach becomes

effective.

S-1

S-2

S-3

3rd report2nd report

Figure 1: An example of stagnation symptom in a document

group.

Stagnation symptoms are categorized into the fol-

lowing three kinds:

• One related to the fundamental aspects of project

progress detected from one document. (cf. S-1 in

Figure 1)

• One detected by comparing documents with the

most recent document. (cf. S-2 in Figure 1)

• One detected from changes to a working object in

a series of documents. (cf. S-3 in Figure 1)

In this research, the fundamental portion is deﬁned

as an important portion that is a fundamental part of

a project’s progress. On the other hand, the detailed

portion is deﬁned as a portion that describes funda-

mental contents in detail.

2.2 Stagnation Symptoms Extraction

In extracting the stagnation symptoms listed in the

preceding section, the following problems occur:

• It is difﬁcult to extract stagnation symptoms if doc-

uments are simply compared and analyzed using

plain texts.

• It is necessary to judge whether a detailed portion is

related to the fundamental aspects of the project’s

progress.

• It is not possible to use a method that uses a past

failure case to extract stagnation symptoms if there

is no failure case.

Therefore, we think that progress of a project

should be modeled to solve problems listed above.

This model is named the “progress model”. Stag-

nation symptoms are extracted by analyzing the

progress of the project with this model. The following

functions are given for this progress model:

• A multilevel layer model using a degree of detail

• Situational analysis using color and relational

analysis of details and fundamentals using color

propagation

• The extraction of stagnation symptoms using rules

Thus, stagnation symptoms are extracted by the

ﬂow shown in Figure 2. First, “label tags” that show

attribute information such as work content, ID, and

pointers to a related sentence are added to each sen-

tence for each meaning within an input document. A

“progress model”, which is a structural chart that ex-

presses the progress of a project, is generated from a

document with label tags. At this time, a target doc-

ument with label tags and the most recent progress

model are input. Adding information in the input doc-

ument to the most recent progress model generates a

new progress model. When the project starts, obvi-

ously the most recent progress model doesn’t yet ex-

ist. Instead, a model generated beforehand is input.

Next, we apply “color propagation”, which analyzes

colors used in the model to understand how a detailed

portion inﬂuences fundamental portion. Finally, we

extract stagnation symptoms by applying three stag-

nation symptom extraction rules to a progress model

that has propagated color.

Extraction of

problem symptom

report

…

report

…

Original document

…

report

…

report

The most recent progress model

Extraction rulesExtraction rules

A progress model is

generated by adding

information in the

input document to the

most recent progress

model.

In order to analyze how

a detailed portion in a

model influences core

portion, color of nodes

are propagated.

Giving label tags

Figure 2: Flow of extracting stagnation symptom.

2.3 Deﬁnition of a Progress Model

In this model, the following two kinds of nodes are

used to distinguish fundamental or detailed descrip-

tions and to indicate project progress:

• Status node

A fundamental node that shows the work procedure

of a project that is given prior to a progress model

being generated. This node is drawn in a rectangle,

as shown in Figure 2.

• Label node

This is a node that shows the detailed work for a

ICSOFT 2006 - INTERNATIONAL CONFERENCE ON SOFTWARE AND DATA TECHNOLOGIES

166

status node. This node is added by referring to a

sentence of an input document. This node is drawn

in a circle, as shown in Figure 2.

The status node is a superior position node indicat-

ing the fundamental portion of project. On the other

hand, a label node is given as a subordinate position

node of a status node.

Steps in a project progress such as “research pur-

pose” and “approach” are written in the rectangle of

a status node. The sentence number of an input sen-

tence is written in the single circle of a label node. In

addition, a label node to a current work is drawn with

a double circle.

Labels are painted in four different colors ac-

cording to a content of the target sentences, for in-

stance, “red” is remarkable stagnation in the project’s

progress. Other colors are “white” as not-processed,

“green” as no stagnation, and “yellow” as more in-

spection.

The model has two axes: The “description level

axis”, which is vertical and the “time axis”, which is

horizontal. A “description level axis” shows that near

a status node is fundamental and distant from a status

node is detail. Figure 3 shows a graphical representa-

tion of nodes.

Approach

status node

label node

level

working node

time

detail

core

Figure 3: Example of writing status node and label node.

If some processes are worked in parallel, they are

shown by two lines as an “AND connection”. On the

other hand, if some processes are worked in an alter-

native manner, they are shown by one line

2.4 An Extraction Method that Uses

Stagnation Symptom Extraction

Rule

After generating a progress model and propagating

color, stagnation symptoms are extracted. There are

three kinds of stagnation symptoms as described in

Section 2.1. In order to extract these stagnation symp-

toms, the following three rules are used:

2.4.1 A Rule that Extracts Stagnation

Symptoms Related to the Fundamental

Elements of Project’s Progress

This stagnation symptom is extracted by analyzing

one progress model. Speciﬁcally, if the color of a sta-

tus node connected with a series or an “AND connec-

tion” in a target progress model is red, it is thought

that stagnation symptoms related to the fundamental

elements of project’s progress exist, as shown in Fig-

ure 4. Thus, a sentence in a label node that causes a

status node to be red using color propagation is ex-

tracted as a stagnation symptom. This extraction rule

is deﬁned as rule 1.

Figure 4: Example of stagnation symptom by rule 1.

2.4.2 A Rule that Extracts Stagnation Symptoms

Detected By Comparing the Most Recent

Document with a Target Document

This stagnation symptom is extracted by comparing

the color of a target progress model’s status node with

the color of the most recent progress model’s status

node. There are two cases. The ﬁrst case is an oc-

casion when progress is not seen, such as from white

to white or yellow to yellow. The second case is an

occasion when state has deteriorated, such as from

green to yellow. We think that stagnation symptoms

exist at the portion that corresponds to these cases’ as

shown in Figure 5. Thus, a sentence in a label node

that causes the color deterioration of a status node is

extracted as a stagnation symptom. This extraction

rule is deﬁned as rule 2.

recent progress model

target progress model

Figure 5: Example of stagnation symptom by rule 2.

2.4.3 A Rule that Extracts Stagnation

Symptoms Detected From the Changes of

a Working Target In a Series of Documents

This stagnation symptom is extracted by seeing the

position and situation of a working node in a target

A DETECTION METHOD OF STAGNATION SYMPTOMS BY USING PROJECT PROGRESS MODELS

GENERATED FROM PROJECT REPORTS

167

progress model and past progress models. Specif-

ically, when no progress can be seen in a working

node in a past progress model, it becomes a stagna-

tion symptom as shown in Figure 6. So, in a target

progress model, a case where a working node is con-

nected to the lower side or right side of working node

in a past progress model is a target of this rule. If the

color of the working node is any color except green,

a sentence in the last working node is extracted as a

stagnation symptom. This extraction rule is deﬁned

as rule 3.

recent progress model target progress model

2 222

3 3

Figure 6: Example of stagnation symptom by rule 3.

3 EXPERIMENTAL RESULTS

We applied the proposed method to an actual failure

case in a research project at university, and extracted

stagnation symptoms. The data of the input docu-

ments used in the experiment are as follows:

• Kind of document: Regular progress report

• Style of report: One A4 size paper

• Volume: ﬁve sheets

In this experiment, stagnation symptoms extracted

with the proposed method are compared with ones

taken by hand. Figure 7 shows a generated progress

model and portions where stagnation symptoms were

extracted.

Manual and the proposed

method could extract.

Manual couldn’t extract though

the proposed method could extract.

Manual didn’t extract though

the proposed method extract

Research

purpose

Research

background

Approach

Basic

technique

Figure 7: The extracted stagnation symptoms from the

progress model.

Some of extracted sentences are shown below.

• Manual operation and the proposed method ex-

tracted the following elements:

– “It is necessary to design other approaches.”

Portion: The fourth sheet, Rule: rule-1

• The proposed method extracted the following ele-

ments that manual operation could not extract:

– “Differentiation with existing research”

Portion: The third sheet, Rule: rule-3

• The proposed method extracted the following ele-

ments that manual operation didn’t extract:

– “There is repetition with the product trend key

word”

Portion: The fourth sheet, Rule: rule-2

The proposed method could extract three stagna-

tion symptoms that the manual operation couldn’t.

These were the stagnation symptoms detected from

situation changes in the past, so the manual extrac-

tor had overlooked these stagnation symptoms. A

portion that the extractor judged weren’t stagnation

symptoms were extracted by the proposed method.

4 CONCLUSION

From the results of the evaluation experiments, we be-

lieve that it is possible to extract stagnation symptoms

from actual progress reports.

The following areas are our future work:

• Automatically adding label tags to the input docu-

ment data.

• Introducing the importance degrees of the stagna-

tion symptom

As for the automatic addition of the label tags, if

input progress reports are assumed to made with a

predeﬁned format, we think that the label tag can be

added by reading words and analyzing the layered

structure of the sentences.

The proposed method ﬁgures out whether there are

stagnation symptoms or not. However, there are dif-

ferent degrees of stagnation symptoms. Thus, we

think that we need degree of importance to judge the

extracted stagnation symptoms as numerical values.

REFERENCES

Kloppenborg, T. and Petrick, J. (2004). Managing project

quality. In Engineering Management Review, IEEE

Volume 32, pp.86 - 90.

Paul, R.A, K. T. S. Y. and Khan, M. (1999). Software

metrics knowledge and databases for project manage-

ment. In Knowledge and Data Engineering, IEEE

Transactions on Volume 11, pp.255 - 264.

ICSOFT 2006 - INTERNATIONAL CONFERENCE ON SOFTWARE AND DATA TECHNOLOGIES

168

Sakthivel, T. S. and Kalyanaraman, V. (1993). Standards

processing in an integrated engineering system. In

Computing Systems in Engineering Volume 4, pp.295-

305.

Takemoto, S. (2005). Study for project management using

practical evm. In IPA Forum 2005 SEC conference.

Tsukuda, G. and Morita, Y. (2004). Using project status

reports to create and reuse project management know-

how. In in Proc. of 2nd International Conference on

Project Management, pp. 72-77.

Y. Uchida, G. Tsukuda, K. H. and Tatebe, K. (2005). Pro-

posal of a risk management support method using fail-

ure knowledge. In in Proc. of The Sino-Japan Work-

shop on Meta-synthesis and Creativity Support System

(CD-ROM).

A DETECTION METHOD OF STAGNATION SYMPTOMS BY USING PROJECT PROGRESS MODELS

GENERATED FROM PROJECT REPORTS

169